University of Pennsylvania Carey Law School Penn Law: Legal Scholarship Repository
Faculty Scholarship at Penn Law
1997
Dividends, Noncontractibility, and Corporate Law
William W. Bratton University of Pennsylvania Carey Law School
Follow this and additional works at: https://scholarship.law.upenn.edu/faculty_scholarship
Part of the Business Administration, Management, and Operations Commons, Business Law, Public Responsibility, and Ethics Commons, Business Organizations Law Commons, Economic Theory Commons, Jurisprudence Commons, Law and Economics Commons, Securities Law Commons, and the Work, Economy and Organizations Commons
Repository Citation Bratton, William W., "Dividends, Noncontractibility, and Corporate Law" (1997). Faculty Scholarship at Penn Law. 880. https://scholarship.law.upenn.edu/faculty_scholarship/880
This Article is brought to you for free and open access by Penn Law: Legal Scholarship Repository. It has been accepted for inclusion in Faculty Scholarship at Penn Law by an authorized administrator of Penn Law: Legal Scholarship Repository. For more information, please contact [email protected]. DIVIDENDS, NONCONTRACTIBILITY, AND CORPORATE LAW
William W. Bratton*
INTRODUCTION It is the custom to hold out Warren Buffett and Berkshire Hathaway as a real world reproach to the arrogance of advocates of financial economic theory. I am in sympathy with this custom. But this Article will make no such reproach, even though it consid ers a practice of Buffett and Berkshire in the context of a body of financial economics. The subject practice is Berkshire's anomalous policy of paying no dividends.1 The subject economic models, which will be collectively referred to as the "first principles variant of incomplete contracts theory," apply the theory of incomplete contracts to the problem of optimal capital structure. In so doing, these models create no occasion for a real world reproach. They show that financial economics, like Berkshire Hathaway itself, has grown and become more complex as the years have passed. Gone are the first-best certainties and simplifications characteristic of first-generation blockbusters like the capital asset pricing model, the efficient capital market hypothesis, and the irrelevance theory of capital structure. These second-generation exercises remit us to a second-best world-a world that, although highly stylized, would at least be recognizable to Benjamin Graham and David Dodd.2 Some strong parallels to the first-generation economics of op timal capital structure nonetheless persist in this second-genera tion, second-best world. Here, as there, debt solves certain governance problems attending equity control, and the issuance of equity in turn solves certain problems attending the incurrence of debt. But here, unlike there, conditions of uncertainty render un-
·· Professor of Law and Governor Woodrow Wilson Scholar. Rutgers School of Law Newark. My thanks to Joe McCahery. David Carlson, and Dale Oesterle for their com ments on earlier drafts of this paper. Special thanks to my research assistant. Andrew \Vhite. I Be rkshire Hathaway has paid only one dividend since Warren Buffett gained control ol it in 1965. That diviclencl was I 0¢ a -.hare distributed in 1967. Later Buffett said that "h e must have been in the bathroom·· when the Board made the declaration. See RoGER. Low E01STEIN. BuFFETT: THE iVI."-.KI>-:G OF ,\N A;viFR.ICAN C.A.PIT..\LIST 130. 133 n.•:• ( 1995). 2 Benjamin Graham was Warren Buffett's mentor. See id. at 36-59. Gra ham coauthored a famous text on securi ty analysis with David Dodd. See genernllv BENJ.-\;..IJN GRAHA\1 ET AL.. SECURITY ANALYSIS: PRINCIPLES ,\NO TECHNIQUE (4th eel. 1962).
409 410 CARDOZO LAW REVIEW [Vol. 19:409 known and unknowable the precise and practical measure of the optimal mix of the two. The models described here reach this com mon-sensical result because they follow an economic theory of the firm that alters a number of assumptions made in first-generation economic models of agency relationships. The first-generation models remitted governance problems to corporate actors for con tractual solution, in many cases undaunted by the apparent ab sence of contractual technologies for dealing effectively with the problems they identified. They assumed, not unreasonably but perhaps not so safely, that suboptimal institutional conditions would create incentives to spur the development of new contrac tual solutions. And they deemed that the future would bring any necessary technical innovations, provided that no regulatory stum bling blocks cropped up to impede this progressive evolutionary process. In contrast, the incomplete contracts models suggest that in formation asymmetries-in particular problems of ex post observa tion and verification-structurally delimit the class subject matter suited to travel on this track of evolutionary improvement. This body of theory remits us to a second-best world for the purpose of identifying and explaining barriers that prevent the evolution of first-best corporate governance institutions. This does not negative the proposition that state intervention can be one such barrier.3 But, at the same time, the state's removal of itself does not neces sarily free transacting actors to cause institutions to evolve to the first-best ideal. It instead holds, first, that transacting actors can work such marketplace magic only to the extent that their subject matter is contractible. Second, it holds that contractibility cannot safely be assumed-the requisite transactional technologies may not yet exist; nor may they even be imaginable in the present state of things. And, third, it holds that corporate capital structure presents many such problems of noncontractibility. It accordingly predicts that the mandatory and contractual devices that vest and transfer corporate control will continue to constitute the central governance institution. By default, then, state intervention retains a place on this theory's list of possible means to the end of improv ing suboptimal governance conditions.4
3 For this view of history. see MARK J. RoE. STRONG MANAGERS. WEAK OvmERs: THE PouTIC";\L RooTs OF AiYIERICAN CoRPORATE fiNANCE (1994). 4 Sec Philippe Aghion & Benjamin Hermalin, Legal Resrricrions on Privaic Conrracts Con Enhance Efficiency. 6 J.L. EcoN. & 0RG. 381 (1990'1. 1997] DIVIDENDS 411
Thus postured, these incomplete contracts models offer no present template for an optimal real world governance regime even though they direct themselves to the business of articulating formu lae for optimal capital structures. To bring them to bear on divi dend and reinvestment policy, then, promises no wealth maximizing quick fix-contractual, mandatory, or otherwise. A pair of more limited objectives must suffice for this Article. First, the models will be used here to explain why dividend and reinvest ment policy has a history of chronic insusceptibility to easy regula tory improvement-contractual, mandatory, or otherwise. Second, the models will be used to appraise the three items on the standing menu of governance reform proposals respecting dividend and re investment policy-specifically, mandatory payout of earnings, in stitutional investor monitoring, and stepped-up disclosure requirements. This Article has three parts. Part I examines the dividend pol icy both of Berkshire Hathaway and of the companies in which it presently holds substantial common stock investments. It there turns out that the apparent puzzle presented by Berkshire's prac tice of total earnings retention is quickly solved with a reference to Graham and Dodd's classic work on security analysis. But it also turns out that Berkshire holds significant blocks of stock in firms that follow a more conventional payout pattern. A puzzle is en countered at that point, but not a puzzle usually connected to Berkshire Hathaway. It is instead the famous dividend puzzle of financial economics, along with the agency explanation favored in legal theory. Part II describes the approach to capital structure emerging in the incomplete contracts literature. The models teach, first, that intractable informational asymmetries prevent direct contractual solutions to the governance problem presented by dividend policy, and, second, that solutions can be structured only indirectly through the control transfer provisions built into corporate capital structures. This story echoes that of the standing agency explana tion of dividend policy, remitting attention to the disciplinary properties of debt for a means to counter the empire-building ten dencies of corporate managers. Unlike the agency explanation, this story does not purport to offer a complete solution to the prob lem of suboptimal earnings retention. It does, however, provide a powerful explanation for the continuing absence of a first-best so lution. Given conditions of uncertainty, it follows from the nature ' 1
412 CARDOZO LAW REVIEW [Vol. 19:409 of debt and equity that the precise measure of an optimal mix of two will remain unknown. Part III uses the incomplete contracts perspective to appraise three legal strategies for ameliorating the problem of suboptimal dividend and reinvestment policy. First, the recent proposal of a mandatory shareholder option to require payout of a pro rata share of earnings is examined.5 Incomplete contracts ideas expbin this strategy's intuitive appeal while simultaneously warning of sig nificant perverse effects. Second, the indirect solution to the prob lem promised by the proponents of institutional investor activism-high-intensity boardroom monitoring by genuinely in dependent directors, is considered.6 Although this strategy proves consonant with the incomplete contracts description of the firm, it remains hobbled by the problem of real world feasibility. The third strategy is stepped-up disclosure-a mandate to management to describe particulars respecting dividend and reinvestment deci sions.7 Here, the incomplete contracts literature sends an equivo cal signal. It supports no prediction that management would respond to such a mandate with credible reports. But it also im plies that revised disclosure rules should not be dismissed out of hand. Although cheap talk has a low value, that value still may suffice to make a disclosure mandate cost beneficial.
l. BERKSHIRE HATHAWAY AND THE DIVIDEND PUZZLE
A. The Berkshire Hathaway Anomaly
Most large, mature American corporations maintain a low but steady dividend payout. Berkshire Hathaway's policy of paying no dividends thus sets it apart. But this anomaly readily can be ex plained. It results from the coincidence of Mr. Warren Buffett's personal preferences and his views on management policy. On the personal side, Buffett has derived life long satisfaction from wealth accumulation.S On the policy side, Buffett respects the bottom line rule respecting the reinvestment of earnings enunciated in corpo rate finance textbooks. Buffett has enunciated it himself:
5 Sere Zohar Goshen. Shareholder Dividend Op1ions. 104 Y'\LE L.J. 88!. 903-06 (! 995 ). r, See inji·a text accompanying notes 141-62. 7 See Victor A. Brudney, Dividends, Discrelion, and Disclosure. 66 Vr\. L. R1·:v. 8.) (I 080): Aaron S. Edlin & Joseph E. Stiglitz. Discouraging Rivals: i'vlanagerial Rem-Seeking und Economic Inefficiencies. 85 A�i. EcoN. REv. 1301 (1095). Bw see Daniel R. Fischel. Tile Lmr and Econo111ics of Dividend Policy, 67 VA. L. REv. 699 ( 1 Y8l ). :-; See Lmn:NSTEIN, supra note l, at 10. 22. 87-88. 1997] DIVIDENDS 413
Unrestricted earnings should be retained only when there is a reasonable prospect-backed preferably by historical evidence or, when appropriate, by a thoughtful analysis of the future that for every dollar retained by the corporation, at least one dol lar of market will be created for owners. This will happen only if the capital retained produces incremental earnings equal to, or above, those generally available to investors.9 Restating the point, corporate cash flowsshould be reinvested only if management has a proj ect that promises a rate of return r greater than the cost of its equity capital k. to The interest of the present case lies in the fact that Berkshire pays no dividends even as Buffett intones the textbook truth with complete plausibility. Buffett legitimately can preach this sermon despite his payout record because he has been the corporate com munity's premier reinvestor of earnings during the thirty years since Berkshire last mailed its shareholders a check. 11 He also is one of the few prominent investment managers ever to liquidate an investment company in a bull market on the ground that advancing prices had eliminated the set of attractive opportunities.12 So when Buffett talks reinvestment policy, the talk is not as cheap as usual.13 His track record amounts to a plausible reputational bond. Although an empire builder, he also cares about his record for making investments with outstanding yields, and, when given the choice, demonstrably prefers to enhance his reputation for yields.
'.! Lawrence A. Cunningham. Compilation. Tlze Essavs of Warren Buffeil: Lessons for Corporare America. 19 C\Roozo L. REv. l. 124 (1997) [hereinafter Buffer! Essays}.
1° For a textbook rendition of this point. see VICTOR BRUDNEY & WILLf,\\·I W. B RAT· TON. CoRPORATE FINA'lCE: CAsEs AND MATERit\LS 548-49 (4th eel. 1993). It is noted that both Buffett and Charles Munger object to the text's association of their opportunity cost rule of thumb for earningsre tention with the financial economic concept of cost of ca pital. They object to the invocation of the latter concept on the ground that it is not subject to empirical verification. See Lawrence A. Cunningham. Editor. Conversariom ji·o111 rhe War ren Bufferr Symposiu111. 19 CARDozo L. RE v . 719. 769-75 (1997) [hereinafter Buffeu Con verso rions J. I I See supra note I. 12 That was the original Buffett Pmtnership. a private investment company managed by Buffett and li qu idated in 1969. See LowENSTiciN. supm note J. at 11-1-15. The recent disso lutions of some well-known hedge funds present a contemporarv variation on the theme. Se!' Laura Jereski. Od1·ssn Dissoh·es 53 Billion Finn: Big Hedge-Fund Era Aiuv Be over. 'vV.\LL S1. J . . Jan. l3. 1997. at B-1. 1 3 All the samt�. some of Buffett's less successful investments-LiSAir and Champion International. made in i9f\9 contemporaneously wi th a successful investmen t in Gillette have been accounted for as resulting iargelv from the fact that Berkshire had cash to in vest. Sa LowE'-!STFI�;. supro note ! , at .15-1-55. 414 CARDOZO LAW REVIEW [Vol. 19:409
Thus, his preferences largely coincide with those of his outside investors. 14 So far, then, Berkshire presents an anomalous pattern of divi dend and reinvestment, but no dividend "puzzle." Perhaps a puz zle will emerge upon the comparison of Berkshire's corporate level policy of total retention with its corporate level treatment of the payout patterns of subsidiary companies. With the subsidiaries, Buffett has been said to do a complete volte fa ce. The story is that, as hard as nails, he resolves all doubts in favor of low levels of earnings retention, and insists that all available subsidiary cash flows be forwarded to the parent company's coffers. Following a practice that started when he took over Berkshire's textile mill, he keeps a sharp eye on every capital outlay in his empire, right down to the office pencil sharpener.15 But, assuming that there is an element of truth to this story, we once again need look no further than the Graham and Dodd textbook16 for an explanation. Graham and Dodd counseled that growth stocks should be valued based on earnings alone, but that with low-growth and declining companies, both dividends and earnings should be considered. More specifically, they recom mended that below average shares in declining companies should be valued in accordance with a formula that accords four times the weight to a dollar paid out as a dividend than the weight accorded to a dollar retained in the business:
Value = Multiplier (Expected Dividend + 113 Expected Earnings) 17 Shares of the large group of low-growth companies falling in be tween growth companies and declining companies should be val ued, said Graham and Dodd, on an intermediate basis, somewhere between the 1 to 1 ratio for growth companies and the 4 to 1 ratio
14 If the dividend payment pattern prevailing in other companies evolved into a stan dard practice as a check against management's tendency toward slack investment practice. then it is easy to assume that an exception would be carved out for Buffett. See infi'o text accompanying notes 37-41.
15 See LuwF:NSTEIN. supra note L at 136-37. Buffett takes issue with the extension of this description to today's Berkshire. He tells us that where Berkshire controls a subsidi ary. but a minority block of stock remains outstanding. the dividend decision is left to those minority shareholders. Sec Buffcrr Conversorions. supra note 10. at 769-75. However. an inference or an alignment or interests in a high payout rate still arises.
lh See GR!\I-IAM ET AL. supra note 2.
'7 !d. at 513. The formula nu1kes it look as if a declining company can retain up to one third of its earnings without impairing shareholder value. but the appearance is deceiving. All other things being equaL the higher the pavout. the higher the value. 1997] DIVIDENDS 415 for decliners.18 Presumably, one would adjust the formula for such an intermediate company after an appraisal of the capital needed to keep its operation up, running, and competitive in its established market. Or, to draw again on Buffett: In many businesses-particularly those that have high asset/ profit ratios-inflation causes some or all of the reported earn ings to become ersatz. The ersatz portion-let's call these earn ings "restricted"-cannot, if the business is to retain its economic position, be distributed as dividends. Were these earning� to be paid out, the business would lose ground in one or more of the following areas: its ability to maintain its unit volume of sales, its long-term competitive position, its financial strength. 19 If the variegated portfolio of going concerns accumulated within Berkshire Hathaway now is examined, any puzzle respecting Berkshire's policy of high internal dividends is quickly solved. Berkshire collects established, well-managed businesses with se cure positions in stable markets-toll booths in the ordinary path of the average consumer, like local newspapers without competi tors and local retailers with successful formulas and large market shares.20 These revenue spinners need to be fed diets of "re stricted" earnings so that they can maintain their franchises, but that is all. The remaining cash flows are released to the parent. There Buffett continues the hunt for new investments in like-posi tioned money spinners. A meaningful survey of Berkshire's dividend and reinvestment policy must include a third class of dividend and reinvestment deci sions. These decisions are made by the public companies in which Berkshire holds large equity blocks and often takes a seat on the board-the famous group of Amex, Cap Cities (now Disney), Coke, Gillette, Sally, and Wells Fargo. This Article's Appendix sets out data respecting the dividend payout ratio for each of these companies, comparing its recent practice to those of its industry group. In some cases the Berkshire-held company pays out slightly more than its industry average, in some cases slightly less, and in some cases about the same. The numbers show at a minimum that Buffett does not use his influence in these boardrooms to effect dividend policies replicating those he has been said to deem sound
1x 5)ee id. at 516-18. I
B. Th e Dividend Puzzle and the Agency Solution Large American corporations shape their dividend policy to accord with a conventional wisdom. Under this, the payout level should be set as a fixedamount rather than as a fixedper centage of earnings yielding a fluctuating amount. Increases should be ap proved only once the new, higher payout level clearly can be sus tained against negative shocks to corporate cash fiow.26 In the
1 2 Or. if he attempts to use his inAuence. he does not succeed in persuading these firms to change their policies. 22 See LowENSTEIN. supra note 1. at 201. 23 See I.R.C. * 243(a) (West 1997) (allowing corporations to deduct 70°/c, or 100% of c!iviclends received. depending on the extent of ownership interest in the payor firm). 24 See John Lintner. Dis1ribu1ion of lnco111es of Corpor{f[ions Al/long Dividends, !<.e tained Eamings. and Taxes. 46 Ai\I. Eco:--�. Rev. 97 (1956). 25 Buff'e/1 Essavs. supro note 9. at 127. 26 Empirical confirmation is set out in Lintner. supm note 24. For a recent reconf1rrna tion of the value of Lintner's model in understanding stock price behavior. see Hyun Mo Sung & Jorge L. Urrutia. Long-Term and Shorr-Term Causal Relarions Benvecn Dividends 1997] DIVIDENDS 417 event of such a shock, the firm should (if possible) borrow to main tain the dividend until relief comes in the form of a cyclical recov ery. A second conventional wisdom about dividends prevails in American textbooks on finance. As noted above, corporate cash flows should be reinvested only if management has a proj ect that promises a rate of return r greater than the cost of its equity capital k. Absent such a project, management should pay out earnings as dividends, or, in the alternative, devote earnings to repurchases of the corporation's shares. These two conventional wisdoms have not synchronized well in practice. The first has synchronized even less well with the basic precepts of financial economics. The divi dend puzzle arises from this dissonance. Financial economics holds at its theoretical base point that div idend policy is irrelevant, at least in a taxless world. This is the famous l.Vlodigliani-Miller proposition: So long as r > k respecting all corporate investments, capital is just as well held in the firm as paid out as a dividend; shareholders desiring periodic cash returns can make their own dividends by liquidating a portion of their stockholdingY However, relaxing the assumption of a taxless world causes the prediction to change. Given the Internal Reve nue Code, we should see a pronounced bias against dividends be cause they are taxable at ordinary income rates, where retained earnings import a tax deferral and a downward shift to capital gains rates. Assuming r > k investments, then, a dividend injures a tax paying shareholder. The prescription changes slightly for firms with free cash flow-that is, internally generated cash in excess of the cost of the set of r > k investments. These monies should be paid out of the firm, but the taxpaying shareholder with a long term holding perspective will prefer a share repurchase program to a dividend. Combining these points yields a rule of thumb respect ing sources of capital for new r > k investments. The first choice is retained earnings, since they carry the lowest transaction costs, and retention avoids the taxable event of a dividend payment. The sec ond choice is debt, since interest payments can be deducted as a business expense. New equity comes in last.2i) Despite this analysis, managers follow the convention of pay·· ing steady dividends. They perceive that any departure from this
und Srock Prices: A Tesr of Linlner"s Dividend ;\llodel and rlze Prescnl Value /Vlodel ofSwck Prices. J8 J. fiN. REs. 171 (1995). 27 Sec Menon H. Miller & Franco Modigliani. Dividend Policy, Grow1!z. and !lzr> Valua lion of Shares. 34 J. Bus. 411 ( 1961). :>.x For a theoretical model of this '·pecking order'' proposition. sec Stewart C. iV!yers. Tlze Capita! s·rruuur!' Puzzle. 39 J. FIN. 575 (1984). 418 CARDOZO LAW REVIEW [Vol. 19:409 practice would go against shareholder preferences with destabi lizing results, even a departure occasioned by an especially good investment opportunity. And one of the better-established empiri cal propositions in financial economics offers them indirect sup port: Stock prices go up when firms announce dividend increases and decline when they announce reductions.29 The dividend puzzle lies in the tension between the practice of steady payout and the theoretical instruction that payout policy should yield to good in vestment opportunities.30 Financial economists have endeavored for three decades to provide rational explanations for the payout practice, thereby solv ing the puzzle. Two leading (and rival) schools of explanation have arisen, and, like conventional dividend policy itself, have persisted. According to one line of thinking, the "signalling" explanation, div idends ameliorate information asymmetries-firms pay them to signal private information about firm profitability.31 According to the other, "agency" explanation, stable dividend policy palliates management's tendency to reinvest free cash flows in suboptimal r < k investments.32 Advocates of both explanations can point to supportive empirical studies.33 Meanwhile, corporate law com-
2
suggest that the marginal cost of dividends is high. If the agency hypothesis is correct this will not be the case. Their study empirically proves the assertion, strengthening the case for the signalling hypothesis. See id. at 549. For a study supporting the agency hypothesis, consider Mahmoud A. Moh'd et aL An Investigation of the Dynamic Relationship Between Agency Th eory and Dividend Policy, 30 F!N. REv. 367 (1995). This is a time-series cross-sectional analysis of 341 firms for the period 1972-89. The analysis presents a series of interesting findings: (1) that firms exper iencing or about to experience high rates of revenue growth tend to establish lower divi dend payouts: (2) that dividend payout increases as a function of firm size (supporting the view that larger firms have higher agency costs and smaller firms have higher financing transaction costs): (3) that dividend payout is inversely related to intrinsic business risk: (4) that firms establish a lower dividend payout as their operating and financial leverage mix increases (see infra text accompanying notes 39-40 (supporting Rozeffs view of dividends as a quasi fixed charge)); (5) that higher dividend payouts are observed when management holds a low percentage of shares and as outside ownership becomes more dispersed: and (6) that firms tend to establish higher payouts as institutional ownership increases. See Moh'd et aL supra, at 379-80. "4 1l1is may also follow from the fact that the signalling models rely heavily on mathe matical description. The agency literature tends to be more discursive. An outside reader of literature on both sides of this debate wonders why the partici pants presuppose that one explanation must be adopted at the expense of the other. Sig nificantly, the theories have been deployed along parallel lines. For example, they have been separately drawn on to explain that. despite the tax advantages of share repurchases, the repurchases are not necessarily superior from the point of view of an outside investor. Michael J. Brennan & Anjan V. Thakor, Shareholder Preferences and Dividend Policy . 45 J. FIN. 993 (1990), provides an asymmetric information model that points out that nonpro portionate repurchases potentially transfer wealth from small shareholders, who have no incentive to become informed about market activity. to large shareholders. who do. Com pare Michael J. Barclay & Clifford W. Smith. Jr., Corporare Payow Policy: Cush Dil·irlends versus Open-Marker Repurchases. 22 J. FIN. EcoN. 61. 65 ( 1988). stressing that manage ment discretion to time repurchases dissipates any bonding effect. These two treatments might well be compared with a third, which follows from the Graham and Dodd tradition. See LouiS LowENSTEIN. SENSE ;\ND NoNSENSE IN CoRPO RATE FINANCE 144-76 (1991). 35 See Drew Fudenberg & Jean Tirole. A Th eory of Income and Dividend Snworlzing Based on Incumbency Rents. 103 J. PoL. EcoN. 75. 78 (1995). y, ·n1 e exception in the legal literature is FischeL supra note 7. at 700. 708- 14. which unequivocally rejects the agency explanation. positing that control market uiscipline obvi ates any problem. Fischel also cites the signalling hypothesis 1v ith complete approval. -'7 See Michael C. Jensen. Eclipse of rhe Public Corporation. HARV. Bus. REv., Sept. Oct. J 989. at 6L 66 [hereinafter Jensen. Eclipse]. 420 CARDOZO LAW REVIEW [Vol. 19:409 the risk averse manager's wariness of a stepped-up ratio of debt to equity.38 The convention of a steady dividend payout results in a check against this tendency toward suboptimal investment of inter nally generated capital. For every dollar pumped out as a dividend, the investing manager has to resort to outside capital markets for a new dollar. A convention of a steady payment stream thus forces ongoing reliance on outside financing.39 The dividend thus bonds the managers to act in the shareholders' interest. Said Rozeff, the originator of this agency explanation, the resort to outside funding forces management to reduce agency costs and reveal information to actors in the capital markets.40 Easterbrook, restating and ex tending the theory, stressed the latter point-dividends "start up" monitoring by capital market actors who, unlike shareholders, are unhobbled by collective action problems.41 The agency explanation solves the dividend puzzle, narrowly defined, by telling us why, despite a tax disadvantage, a share holder rationally might prefer one dollar to be paid out as a divi dend rather than reinvested in an r > k project. But, in so doing, it gives rise to a new question: If conventional payout policy has evolved as a solution to the agency problem bound up in manage ment discretion over investment and financing policy, why, despite widespread adherence to the steady payout convention, have divi dend and reinvestment practices widely been perceived to be
3K See Goshen. supm note 5. at 887-88. 3<.J See Easterbrook. supra note 30; Michael S. Rozeff, Growth. Beta and Agency Cosrs as Dererminams of Dividen d Favour Rarios. 5 J. FIN. REs. 249 (1982). 4ll See Rozeff, supm note 39, at 250. The bond, although reputational only. does impact on management behavior. See Goshen. supra note 5. at 890-91. The empirical studie� focus on firms in financial distress. and show that they tend to cut rather than eliminate their dividends. See Harry DeAngelo & Linda DeAngelo. Dividend Policy and Fin ancial Disrress: An Empirical ln vesrig ation of Troubled NYSE Firms. 45 J. fiN. 1415 (1990): Al bert Eddy & Bruce Seifert. Dividend Ch anges of Fin anciallv We ak Firms, 21 fiN. REv. 419 (1986). Rozeff. strictly following the original agency paradigm. also suggests that dividend pol icy and insider ownership are substitute tools to reduce agency costs. with firms with high percentages of insider ownership paying small dividends. See Rozell mpra note 39. at 25 1. Subsequent empirical work negates this picture. Diane K. Schooley & L. Dwayne Barney. Jr.. Using Dividend Policy and Man agerial 01vnership ro Reduce Agency Costs, 17 J. fiN. Res. 363 (1994). show that the relation between the dividend payout ratio and m�mage ment ownership is nonmonotonic. Beyond a certain point. greater management ownership causes the dividend payment to rise. Schooley and Barney conclude that their results ac c'.>rd with the management entrenchment hypothesis of Randall Morek et al., ,"vf anagemenr 01vnership and i'vl arf..:.et Va lumion . 20 J. FI!':. EcoN. 293 (1988). ll1at is. at some point increases in management stock ownership increase agency costs. -+I See Easterbrook. supra note 30. at 653. 655. He adds a point about management risk aversion-the need to replace internal flows may prompt leverage beneficial to sharehold ers that managers otherwise would avoid incurring. See id. at 653-54. 1997] DIVIDENDS 421
suboptimal?42 The remainder of this Article draws on incomplete contracts models of corporate capital structure in an attempt to an swer this question.
II. AGENCY COSTS, CAPITAL STRUCTURE, AND N ONCONTRACTIBILITY TI1e following sections set out basic components of the incom plete contracts model of capital structure and then extend the model to dividend policy. This presentation requires some antece dent contextual grounding, provided in the first two subparts be low. The first shows how these models' assumptional framework differs from that of the earlier and better-known generation of in complete contracts models. The second shows how this framework has been brought to bear on the matter of optimal capital struc ture. The discussion-in-chief begins with the third subpart.
A. Alternative Approaches to Contractual In completeness Transaction Costs and First Principles Incomplete contracts models deal with the problems that arise when contracting parties possess less information than is necessary to approximate their first-best expected utility. Most work on in complete contracts falls into one of two basic paradigms: the origi nal transaction costs approach, and a newer, and contrasting, approach which derives incompleteness from first principles. The transaction costs approach has been assimilated into the con tractarian theory of the firm of law and economics. The compo nent ideas of the first principles approach as of yet have shown up sporadically in legal commentaries.43 First principles models will be applied here. Under the transaction costs paradigm, costs prevent actors from writing complete ex ante contracts and contract-inhibiting costs continue to accumulate throughout a contract's life. Costs in-
42 Goshen. supra note 5, at 887-88 nn.34-35. renews the denunciation but cites studies dating from the 1980s and earlier. It is unclear whether the problem persists in the 1990s with its former intensity. 43 See, e.g., William W. Bratton et al., Repeated Games, Social Norms, and Incomplete Corporare Conrracrs. in FAIRNESS AN D CoNTRACT 163, 166-71 (Christopher Willets ed., 1996): Oliver Hart. An Economis!"s View of Fi duciary Dwy. 43 U. ToRONTO L.J. 299 (1993): Oliver Hart. An Economisr's Perspective on rhe Theory of the Firm , 89 CoLUM. L. REv. 1757 (1989): Avery Katz. When Should an Offe r Srick? The Economics of Promissory Eswppel in Preliminary Negotiations, 105 YALE L.J. 1249. 1278-79 (1996); Alan Schwartz. Relational Contracts in rhe Courts: An Analysis of Inco111plete Agreements and Ju dicial Strategies. 21 J. LEGAL STUD. 271, 272-73 (1992). 422 CARDOZO LAW REVIEW (Vol. 19:409 curred by the time of execution and delivery amount to an initial investment. They result from the difficultyof anticipating and pro viding for all state contingencies respecting a transaction, whether because of their overwhelming number or because of the large in vestment required to underwrite the composition of, and to sup port the process of reaching agreement upon, specific regulations for each future state. After the contract's execution and delivery, each step in the enforcement process, from verifying counterparty performance (or nonperformance) to invoking third party enforce ment mechanisms, entails additional costs. Ex ante projection of such enforcement costs shapes the choice of agreed upon terms, and, in an extreme case, causes an otherwise productive transaction to be foregone for lack of feasibility.44 The transaction costs paradigm recognizes that contracting ac tors cannot be expected to negotiate complete ex ante solutions to all problems. It nevertheless advances the notion that the institu tion of ex ante contracting, broadly conceived, self-sufficientlysup ports efficient transactional relationships. It makes three assertions toward this end. First, actors who risk capital can be ex pected to design ex ante governance structures that minimize the costs of future uncertainty. Second, even though legal decision makers must assist the parties by filling in omitted terms ex post, those terms may be cast from an ex ante time perspective, and, indeed, should be so cast in order to guard against disruption of the parties' allocation of financial risk and to minimize future transac tion costs.45 Third, comes a prediction. Given proper containment of the agencies of state intervention, transacting actors can be ex pected to devise technologies that lower the transaction costs that cause incompleteness, thereby expanding the effective zone of con tractual governance. The first principles paradigm of incomplete contracts begins with the transaction costs paradigm's diagnosis of the causes of contractual incompleteness. But it then brings the notion of in completeness to bear on a more precise conception of "contract." Unlike the transaction costs approach, which tends to include any voluntary economic relation within its notion of the ex ante con tract, the first principles approach restricts the reach of the ex ante contract to cases where actors make explicit specifications about the future. That is, to have "contract'' terms that govern future states, those contingent states must be specified and the future out-
44 See Bratton et a!.. supra note 43. at l66-7 l. 45 See id. 1997] DIVIDENDS 423 comes must be computable. Since some future states of nature clearly are not computable, transacting parties as a result will lack the technology necessary to enable the negotiation and composi tion of a contract term ex ante.46 The first principles paradigm also places a greater stress on the ex ante impact of ex post problems of performance and enforce ment than does the transaction costs approach. Thus, even where an ex ante computation is theoretically feasible, if a party's per formance of that computed future state will not be both observable by the counterparty and verifiable by the enforcing authority. ex ante agreement on that contract term will not be feasible.47 These three factors-computability, observability, and verifiability-in trinsically limit the operation of the institution of the ex ante con tract.4c; Although each factor results in costs, to characterize the three as ''transaction costs" and nothing more trivializes the seri ousness of the productivity problems they cause. As applied to capital structure, the first principles framework asserts, first, that corporate contracts can be expected to omit im portant future variables due to the difficulty or impossibility of ex ante description or ex post observation and verification, and, sec ond, that given these points of noncontractibility, important out-
4A See Luca Anderlini & Leonardo Felli, lncomplere Wriuen Con/racrs: Undescriboble Srmes of Narure. 109 Q.J. EcoN. 1085 (1994). 47 For fundamental contributions to the literature making this point, see Sanford Gross man & Oliver Hart. Th e Cosrs and Benefils of Ownership: A Th eory of Ve nical and Lmeral !111egrmion . 94 J. PoL. EcoN. 69 1 (1986) [hereinafter Grossman & Hart. Ve nica/ & Lo!erul fnregmrion]: Oliver Hart & John Moore. ln comple£e Coll/racls and Reneg01imion. 56 EcoNo:-.tETRtCA 755 ( 1988). 4:-> Process int1rmities also limit the utility of contract. Even where parties could cost heneficially specify a contract term. information asymmetries and strategic behavior may prevent them from doing so. 'The first principles perspective insists that bargaining processes often shape contractual results. and models the problems that come up when relational economic actors transact. 1l1 is leads to the question of whether a viable set of governance provisions for a firm can be derived through any available model of contract. Some bargai ning models show coordination failures. Rational actors can conceivably adopt any one of a number of mutually consistent arrangements and market forces may fail to assure that only efficient pattern s emerge from the range of possibilities. Other models iclentifv costs of bargaining that prevent efficientresults. Consider a price negotiation over the sale of a nonfungible product. A buyer seeking a greater share of the gains of trade might invest in quality information to gain a bargaining advantage. Such an investment in a pure distributional advantage is ineflicient. since only total benefits and costs matter from an efliciency standpoint. See Paul Milgrom & John Roberts. Bargaining Cosls. Influ ence Cosrs. one/ rl1e Orguni�arion of Economic ACiivin·. in PERSPECTIVES o� PostTIVF PoLITtC.-\ L Eco:--J o;o.tY72-77 (James E. Alt & Kenneth A. Shepsle eels .. 1990). In the alter native. each bmgaining party stands to benelit from the communication of in formation about its own prefere nces. The resulting informational uncertainty can produce the loss of a bcncl1cial transaction. and induces inefficient informational investment in any even t. 424 CA RDOZO LAW REVIEW [Vol. 19:409 comes of necessity will be determined by the firm's structure of ownership. The specification of the owner and any associated con tingent control allocations built into the firm's contracts-in partic ular the contracts making up the capital structure-substitute for contract terms absent due to the condition of contractual incom pleteness.49 As the zone of noncontractible contingencies expands, the ownership specifications become more important because the firm'sperf ormance will depend on the incentives of its present and contingent future owners. Notably, ''owner" is here specially de fined as the party who has the right to control all aspects of the asset that have not been given over to contractual specification ex ante.50 Under this definition,ownership and control cannot be sep arated, although they can be shared or transferred. Since asset control is ownership, residual claimants who do not manage are not owners, whatever the law's contemplation.51
B. Debt and the Maximization of Va lue
Since dividend policy is irretrievably tied to investment and financing policy, economic theories of the dividend tend to be cor ollaries to theories of optimal capital structure. The incomplete contracts models follow this pattern. The economic theory of capital structure has evolved as an ex tended response to the famous irrelevance hypothesis of Modi gliani and Miller. Under the Modigliani-Miller model, firm value stems entirely from the production function and is independent of capital structure. Furthermore, the cost of capital is constant across all debt-equity ratios. Subsequent models relax the Modi gliani-Miller assumptions to show dependencies between capital structure and firm value. These variously emphasize debt's signal ling role,52 its role in facilitating monitoring,53 and its commitment
4'l See Philippe Aghion & Patrick Bolton. An !ncomplere Contracts Approach to Finan cial ComraCEing, 59 REv. EcoN. STUD. 473, 479 (1992). 5o See Grossman & Hart. Ve rtical & L(/{eral fmegrarion. supra note 47, at 695. 5! Tbey may have contingent rights to take ownership, or to substitute one owner for another. but so long as they passively receive a payment stream. they are not owners. 52 Sec Stewart C. Myers & Nicholas S. Majluf. Corporare Financing and !nvestmellt Decisions When Firms Have Informarion Th ar !nvesrors Do Nor Have, 13 J. Fn". EcoN. 187 ( 1984 ) ; Stephen A. Ross. The Derenninarion of Financial Srrucwre: The Incentive-Signal ling Approach. 8 BELL J. EcoN. 23 (1977). 5.3 Sec. e.g.. Douglas Gale & Martin Hellwig. fncenrive-Colllparihle Debt Contracts: The One-Period Problem . 52 REv. EcoN. STu o. 647 (1985); Robert M. To wnsend. Optimal Contracrs and Competirive !VIarkers with Cos!lv Slate Ve rification. 21 J. EcoN. THEORY 265 (1979). 1997] DIVIDENDS 425 value,5-+ in each case showing a connection between the firm's debt/ equity mix and the value of its production function. With the inter polation of agency theory the models achieve a tie to management science. Not only is there a theoretical optimal level of debt, but that level will tend to lie higher than the level indicated by the conventional wisdom prevailing among managers due to their pref erence for debt at comfortably low levels.55 This agency analysis extends to dividend policy, with subop timal earnings retention and suboptimal borrowing turning out to be two manifestations of the same agency problem. At its firstap pearance, agency theory also proposed a real world solution Michael Jensen's famous model of a high-leverage firm controlled by a leveraged buy out association.56 This model asserted that the high level of debt attending an LBO bonds management to pay out free cash flowin the form of interest payments and makes it impos sible to reinvest in suboptimal projects. Additionally, the immedi ate threat of bankruptcy in the post-LBO period causes management incentives to be better aligned with those of outside investors. Given a heavy debt load and an equity stake, managers will perform like sole owners. This model, however, lost credibility due to indiscriminate employment in practice during the latter part of the 1980s. Incomplete contracts models of optimal capital structure in the first principles mode revive the point that debt enhances value by controlling agency costs. But in so doing they interpolate a more complex concept of optimality. In earlier models, debt re duced agency costs subject only to the caveat that problems of overinvestment and underinvestment become prohibitively costly when debt reaches extraordinarily high levels.57 The incomplete contracts models look to debt's effects on management incentives across a wider range of decisionmaking scenarios and bring to bear a more extensive range of valuation variables. The problem for solution is not just "cost control" conceived in terms of a short list of types of costs attending states of high debt. Optimal capital structure here also concerns the degree of synchronization between the terms of the debt contract and both the character of the bor-
5-1 See Jensen. Agency Cosrs, supra note 32: Sanford J. Grossman & Oliver D. Hart. Corporore Financial Srrucwre and Managerial lncenrives, in THE EcoNOMICS OF INFORMA TION Ai':D U�CERTA !NTY 108-10 (John J. McCall eel., 1982) [hereinafter Grossman & Hart. Corporure Financial Srruuure ]. 55 See Jensen. Agency Costs, supra note 32. at 323-25. 5(, See Jensen. Eclipse. supra note 37. at 61. 57 For an overview. see BRUDNEY & BRAT!"ON. supra note 10, at 475-77. 426 CA RDOZO LAW REVIEW [Vol. 19:409 rower's asset base and its future sequence of business decisions. We get a multifaceted notion of optimality that focuses as much on the dangers of excess debt as on the agency costs of insufficient debt. 58 A model of short-term debt devised by Oliver Hart provides a good introduction to the idea that a theory of optimal capital struc ture must take account of structural connections between the terms of the debt contract and particulars respecting the borrower's busi ness.59 This is a two period model that assumes complete informa tion. More particularly, investment occurs at t = 0 in a project conceived and managed by an entrepreneur E. The project will throw off cash flowsyl and fy2 at t = 1 and t = 2, where t = 2 is the project termination date. The model assumes that the ultimate agency problem occurs at t = 2: Specifically,E will be positioned to divert all of fy 2 to her own pocket. As a result, if outside financing is to be feasible, provision must be made for complete payment to the outsider at t = 1. Short-term borrowing accordingly emerges as the only feasible vehicle. The model makes a number of additional assumptions. E has all the bargaining power, and the outside investor I who lends the money can be held to a break-even return. In addition, E has a limited amount of wealth w, with w < K, the cost of the proj ect. E must invest w in the project, borrowing at least the difference be tween K and w, with the debt due in its entirety at t = 1. E, having borrowed an amount B, can liquidate project assets in order to make the contractual payment P at t = 1; f is the value of proj ect assets remaining after this liquidation. If E cannot make the pay ment, the firm is liquidated at t = 1 at value L, which goes to I. Thus, the amount I actually receives at t = 1 is equal to Min {P,L}. 6o
50: Compare also in this regard the significant differences between the treatment of bankruptcy in the literature of the 1980s with the treatment today. The agency cost models relied on the assertions that the threat of bankruptcy prompts better management per formance and that bankruptcy process amounts to a low-cost exercise in contract renegoti ation. See Jensen. Eclipse. supra note 37, at 72; Jerold B. Warner, Bankruptcy Cosrs: Some Evirlencl:' . 32 J. FIN. 337 (1977) (direct costs of bankruptcy around five percent). Today. the bankruptcy system is treated as a source of both significant transaction costs and perverse incenti\'es. Significantly. the economists responsible for the i ncom ple te contracting models of capital structure have taken a part in bankruptcy reform discussions. See Philippe Aghion et
Investment Realization of yl Realization ofl\2 K and liquidation of (1 -f ) and liquidation of project I I I I
t=O t= l t = 2
The model defines the parameters of both the class of feasible borrowing transactions and of the optimal debt contract. The leveraged project is feasible if: 1) L :2:: K-w; and 2) E will receive by t = 2 an amount greater than her present wealth w. More partic ularly, assuming that E will have to liquidate some project assets to make P at t = 1, the net present value of the project, yl + fy 2 + (1- f) L must be greater than K. Given these parameters, it turns out the project may not be feasible at t = 0 even though its present value of yl + fy2 > K-w. Feasibility-borrowing capacity-de pends also on the value of L. If K = 90, B = 60, P = 60, yl = 60, fy2 = 70, and L = 30, the proj ect is worth 130 (discounted) - 90, but will not be financed because its liquidation value at t = 1 is only 50 percent of the payment due, and E cannot make a credible com mitment to pay any part of fy2 over to I. If we change the facts so that yl = 50 and L = 60, borrowing becomes feasible but subop timal. E is going to have to liquidate an amount 10 of the project's assets at t = 1 in order to make P, thereby reducing the base of assets in place necessary to produce the fy2 value of 70.61 The model allows for a range of optimal debt contracts. That is, so long as L :2:: B = P :2:: K- w, any amount borrowed is optimal. It thereby teaches relatively little about the properties of the opti mal debt transaction. But some enhancement of the model's heu ristic value results when it is extended to cover multiple periods. Here the model starts with K, w, and B at t = 0, and allows an indefinite number of periods to occur between t = 0 and project termination at date T. The project is assumed to yield an amount y in each one of these intermediate periods, and to possess a differ ent liquidation value L on each intervening date on which an amount y is realized. The project depreciates, so the value of L decreases progressively during the life of the project and L = 0 on the termination date T. The debt contract specifies that payments are due at each date between r = 0 and t = T, with the last payment due on t = T-1, the date immediately preceding T. The formal model of debt capacity that emerges within these extended parameters echoes the conventional wisdom of a Graham
61 See ir!. at I 05-06. 428 CA RDOZO LAW REVIEW [Vol. 19:409 and Dodd bond analyst.62 Debt capacity is stated as a function of the amounts and times of receipt of the flows y and the rate of depreciation of the project's liquidation value. Optimality turns on the relationship between these factors, the amount borrowed B, and the contractual repayment path. Once again, the model allows for a continuum of optimal arrangements. Repayment may occur quickly over time, that is, E borrows as little as possible and repays as quickly as possible; repayment also may occur slowly over time, that is, E borrows as much as possible and pays back as slowly as possible.63 But some parameters emerge to narrow the range of equilibria. First, fast repayment paths are facilitated by projects with front-loaded streams of y, and slow repayment paths are facil itated by longer lived, more durable assets.64 Second, a larger amount w, or owner equity, can support a greater quantum of debt.65 Third, determination of the optimality of a given debt ar rangement depends not only on the amounts y and P but on oppor tunities for their reinvestment. If either E or I has reinvestment opportunities superior to the project during the life of the project, then wealth is enhanced to the extent that the project flows are directed to that party. For example, if the superior opportunities lie with E, while I reinvests at a market rate, then the optimal pay ment path is the slowest-the differential between y and P is capi tal for reinvestment by E.66
62 First. total B outstanding at any time cannot exceed the liquidation value L. Second. total B, net of the sum invested in the project plus the present value of the cash flows produced by the project, must be at least as large as the present value of the debt repay ments. Given the continued assumption that E has all the bargaining power and { must invest on a break-even basis. this implies the following expression, with Po standing for the stream of debt payments and Yo standing for the stream of project cash flows:
r r-1
I P, ::0: (K-w; - I Y, for all t = 1 . . ... T. ,�o
For the first-best to be achieved:
1-}
K-w- I Y, ::; Ll fo r all t = l, . ... T. r-,_1) See id. at 108. n3 See id. at 109- 10. If the depreciation comes quickly during the life of the project ancl the cash flows are encl loacled. E can borrow an additional amount to be put in a savings account to cover the differential between the two during the early life of the project. See id. at 109. 64 See id. at 110-ll. 65 See id. at 111-12. 66 See id. at 110. The introduction of uncertainty also reduces the number of equilib rium results. See id. at 112-15. For a recent example of another line of finance literature that articulates models of optimal debt capital structures keyed to the timing of debt payments. see Hayne E. Leland 1997] DIVIDENDS 429
C. The Contingent Control Model of Cap ital Structure Hart thus accesses the basic terms of the debt contract, but with a narrow model that assumes complete information and cab ins all agency problems of debt and equity in a tightly-definedsce nario of end-period opportunism.67 Obviously, a model that successfully accesses the role of debt in controlling the agency costs of equity will have to relax these constraints. In the real world, information will not be complete ex ante, and opportunism will be a possibility during all periods and will extend to discretionary choices respecting investment and effort level. Thepa rameters of a model of an optimal debt contract, thus expanded, will give rise to agency problems that are unobservable or unverifiable ex post. Accordingly, the modeling exercise will have to proceed in an in complete contracts framework. As noted above, the incomplete contract approach predicts that corporate contracts will omit important future variables that are difficult or impossible to describe initially,68 and seeks to show that the control allocations bound up in the firm's capital structure substitute for this incompleteness.69 Since these control arrange ments also determine the identity of the actors who direct the firm's ongoing management and investment policies, or, in the al ternative, determine whether to sell or liquidate the firm, they sig nificantly impact on the value of the firm. In the incomplete contracts perspective, capital structure optimality depends on con trol arrangements.
1. The Contingent Control Model. The base point for examinations of optimal capital structure in the incomplete contracts framework is Aghion and Bolton's model of contingent control allocation.70 It bears close inspection.
& Klaus Bjerre To ft. Optimal Capiwl Strucrure, Endogenous Bankruptcy, and the Te rm Structure of Credit Spreads, 51 .f. FIN. 987 (1996). T1wy show that the optimal quantum of leverage is lower with short-term borrowing. and that short-term borrowing ameliorates asset substitution problems. 67 He assumes that E will be able to get away with nonpayment in the last period and that the problem can be completely solved with a contract that provides for fi nal repay ment in the penultimate period: if the payment schedule does not solve the problem, then debt financing is not feasible. 6S Grossman & Hart, Ve rrical & Lateral fntegrarion, sup ro note 47. and Hart & Moore. supra note 47. set out a model of vertical integration and ownership based on this assump tion of contractual incompleteness. 6 In Aghion and Bolton's model, once again we get a stylized, two-period picture of the relationship between an entrepreneur E and a venture capitalist I. In this model E has no wealth and needs to finance the entire start up cost K for her project. Once again, the world is full of venture capitalists, but contains only a few en trepreneurs with good projects. E as a result has all the bargaining power: E can make a take it or leave it offer which I will accept so long as the deal promises an expected return of at least K. But the payout will depend on an action a to be taken from amongst the set of feasible actions A by the actor in control after the future realiza tion of a state of nature e. At t = 1, prior to the time for the choice of a, the operation of the business will produce a signal s as to the state of nature e. Returns are realized at t = 2.71 Investment Signal s Realization of K as to e returns I I t=O t= I t=2 action a taken This scenario creates a problem. Returns to E and I both de pend on action a, but are received in different forms. Both E and I are risk neutral as to income. Monetary returns of the project r are payable to I at t = 2, minus amounts of compensation payable to E pursuant to a compensation schedule in the contract concluded by E and I. The compensation arrangement provides a transfer t ;:::: 0, the precise amount of which is a function of s and r. Thus, l's pay ment y = r - t. E also receives significant non-monetary private benefits b, such as reputation. These are neither observable nor verifiable by third parties. The model assumes that the quantum of b is a legitimate part of the overall yield of value from the project. Yields of both r and b will depend on the state of nature e and the choice of a. E's yield is a function of r(a,e) + b(a,e), and l's yield is solely a function of r(a,e) . Since the choice of a can differentially impact r and b, a potential conflictof interest as to the choice of a is wrought into the situation.72 If the state of nature e could be specified ex ante, it might be possible for the contract between E and I to direct the party in control, presumably £, to take the jointly maximizing action a. Unfortunately e is impossible or very costly to describe ex ante, 71 See id. at 475-76. 72 See id. at 476. 1997] DIVIDENDS 431 although the parties will be able to identify e ex post. The model does, however, assume that even though the E-1 contract cannot be made directly contingent on 8, it can be made contingent on the signal s, which is publicly verifiable, although imperfectly corre lated with e. The occurrence of s at t = 1 does not, however, enable the drafting of a complete contract. For even if s were perfectly corre lated with 8, the project still would be too complex to permit an ex ante specification of the optimizing response a to be chosen from the set of possibilities A upon the realization of s. Action a is wholly within the realm of management business judgment and is neither susceptible to direct specification nor to indirect specifica tion through a constellation of negative covenants. Direct specifi cation might be possible in a different case where A entailed a selection between a limited set of identifiable choices such as merger, liquidation, sale of assets, or continuation. But, even given the feasibility of that sort of specification, ex post judicial enforce ment of the contractual directive could still fall short of feasibility if information asymmetries led to problems of third-party verification. The upshot is that the capital structure's allocation of control rights between E and I will determine the choice of a and the op timality (or suboptimality) of the value yielded by the firm.73 The capital structure as set out in the E-1 contract inevitably specifies an allocation of control, which in turn determines which actor has the privilege to chose action a. Control can lie in E or in 1.74 A number of additional assumptions are made. First, there are only two possible future states of nature, 8g and eb. Second, there are only two possible outcomes for s, 0 or 1, with s = 1 mean ing that it is more probable than not that e = eg, and s = 0 meaning that it is more probable than not that 8 = eb· Third, action set A contains only two possible actions, ag and ab. in each of the two states of nature 8g and eb. In state 8g the first-best choice of action is ag = a*(8g) ; and in state eb the first-best choice of action is ab = a*(81,) . Fourth, there are only two possible returns r at t = 2, either 0 or 1. Fifth, the initial contract between E and I may be renegoti- 73 See id. at 476-77. 7-+ The model also allows fur the possibility that control can be exercised jointly. In that case the model assumes that either E and I must agree, or in the event of disagreement, E will make a one-time take it or leave it offer to I as to choice of a: in the event that I refuses the offer cleacllock results ancl both parties have 0 returns. Such a joint control setup means that hold ups arc a possibility in every case. As a result, in this model, joint control always is dominmccl by unilateral or contingent control. See id. at 486. 432 CA RDOZO LAW REVIEW [Vol. 19:409 a ted after the realization of 8, with all the bargaining power lying with E. Sixth, given the specification of first best action ag and ab, the expected returns y and private benefitsb realized by I and E in eg and eb will have the following properties: y8gag + b8gag > y8gah + b8gab y8"a" + b8"a" > y8bag + b8bag And, finally,in order to make the initial investment of K plausible, the probability q of yg + (1 - q)yb > K.75 If a* were contractible, control could be accorded to the desig nee with the greatest expertise respecting the production function (here E) with the contract assuring the selection of the optimal course of action. In the alternative world of noncontractibility presented here, a* in theory could be the result of a round of ex post renegotiation occurring after the realization of s. That is, there would occur a round of Coasean bargaining after t = 1 in which a noncontrolling party benefited by the choice of a* purchases its choice by the controlling party with a side payment.76 2. E Control The model works through the scenarios of E control, I control, and joint control to ascertain the distance between the set of results built in by the incentive structure and first-best set of results. Where E controls, first-best choices of action follow in two classes of cases. The firstis that in which the choice of a* also happens to maximize y, b, and the transfer payment t. Here E's incentives are perfectly aligned with the general maximizing result. 77 In the sec ond class of cases, the first-best result does not follow from the incentive structure, but may be reachable through renegotiation. Assuming eb, this occurs where b(8bag) + t > b(811ab) + t, and y(8bab) > y(8bag)· Recalling that E has all the bargaining power, E will offer to choose a* (here 8bab) if I pays E the sum y(8bab) - y(8bag) , provided of course that b(8bab) + t + y(8bab) - y(8bag) � b(8bag) + t. I can be expected to accept provided that the yield of y(8bag) � K, or in other words, so long as he breaks even. Assuming 8g, this occurs where b(8gab) + t > b(8ga�) + t, and y(8gag) > y(8gab). Since E has all the bargaining power, E will offer to choose a*, (here egag) if I pays E the sum y(8gag) - y(8gab), provided of course that 75 See id. at 477-79. 76 See HART. supra note 59. at 98 (discussing the Aghion-Bolton model). 77 See Agh ion & Bolton, supra note 49, at 480-81. 1997] DIVIDENDS 433 b(8gag) + t + y(8gag) - y(8gab) ?:: b(8gab) + t. I can be expected to accept provided that the yield of y(8gab) ?:: K, the break even figure. Such a renegotiation will not result in every case, however. The model assumes that the return of at least K constitutes a ra tionality constraint for I. Thus, the renegotiation will fail, and the first-bestresult will not be chosen if the value of K is so high that it exceeds the yield on offer in E's renegotiation. Indeed, the very possibility of these situations means that I can be expected to re fuse to invest at t = 0 unless some form of protection against E's opportunism is included in the contract package.78 3. I Control Here, first-best choices of action will follow only where the choice of a that maximizes y happens to be a*, meaning that I's incentives are perfectly aligned with the general maximizing result. Where the choice of a that maximizes y is not first-best there can be room for Pareto improving renegotiation. But, as already seen, renegotiation amounts to the payment of a bribe to the actor in control from the actor disadvantaged by the suboptimal choice of a. The model's assumption of a wealth constraint on E's part sub stantially limits the possibility of renegotiation where I controls. Simply, since b and t constitute E's entire wealth, E lacks the re sources to make the bribe. For I control to assure first-bestresul ts, then, the amount of t has to be set high enough to give E sufficient cash for the bribe. This adjustment creates the same situation as the search for the first-best under E control. As t increases, pro jected investment returns to I fall short of K at some point and I refuses to invest. 79 4. Contingent Control-the Effi ciency Function of Debt The Aghion-Bolton model interpolates the device of contin gent control to solve the problem presented by the misalignment of the incentives of E and I. Two additional assumptions have to be made in order to make the model work, however-that y8gag < y8gah and that b8bab < b8bag . These inequalities align the class of cases in which l can be expected to make a suboptimal choice of a to 8.� states and the class of cases in which E can be expected to make a suboptimal choice of a to eb states. With this alignment, I will make a first-best choice in eb and E will make a first-best 7:; See id. at 480-83. 7 D. Implications fo r the Public Corporation ] . Debt, Executive Employment Contracts, and Dividends in a Wo rld Presenting Problems of Observability and Ve rifiability Although the contingent control model deals with a stylized close corporation, it has some important implications respecting agency problems in publicly traded firms. First, the model implies that a provision for the transfer of control to outside investors holding debt may build a governance disincentive into the firm's capital structure respecting management pursuit of private bene fits. To see this, assume that the contract is drafted so that a pay ment is due on the debt at t = 1, and the amount of the payment to be based on a projection such that E should have the ability to meet the payment in a 8g state, but will not meet the payment in a eb state. The provision for a payment accomplishes a transfer of control, with no attendant problems of observability and ver ifiability. Debt's role in agency cost reduction is thus recharacter ized. In earlier agency cost literature, debt bonded E to pay money out to !. Here, it effects a transfer of control to a party better posi� tioned to maximize the value of the assets based on transparent and verifiable events-payment and nonpayment. Note also that the critical event of default need not be a payment default. So long as s is verifiable, it can be employed as the default contingency with payment on the debt being delayed until t = 2. A subsidiary implication becomes apparent if the contingent control model is considered together with a Hart observation re- :->ll See id. at 4S4-S6. :-; 1 The precedent model is found in Jamie F. Zender. Op rimal Financial Instrumenrs. 26 .f. F:�. 1645 (1991). ,-; ;: Set: Aghion & Bolton. supra note 49. at 487. 1997] DIVIDENDS 435 specting the timing of debt payments: Short-term debt is the harder claim from a governance point of view. This point is formalized in subsequent work by Berglof and von Thadden.83 They model a firm which, unlike the firm in Aghion and Bolton's model, has un verifiableincome streams in addition to firm specificassets. Such a firm has difficulty making a credible commitment to lenders ex ante because its position of informational superiority creates op portunities for strategic renegotiation of the debt contract in the event of distress. Themodel shows that a capital structure contain ing secured short-term and subordinated long-term debt held by separate actors will be superior to a structure where the same ac tors hold both short- and long-term claims. It is superior because the stand alone short-term creditor rationally takes a tough posi tion in the event of distress. Since this actor is positioned to fore close on assets and has no long-term interest in the firm, he remains relatively immune to the debtor's strategic renegotiation offer.84 A typology showing variations in the enforcement postures of different securities by degrees of hardness is invited, with short term debt being the toughest, long term debt taking an intermedi ate position, and equity emerging as the soft outside claim.85 Third, the contingent claim model implies a question respect ing the relative effectiveness of employment contracts and control transfer structures as a means to channel management incentives in productive directions. To the extent that crucial management choices-selections of a from sets A-are noncontractible due to problems of observability and verifiability, monetary incentive schemes based on firm profitability or stock market performance cannot be expected to import adequate discipline. Control struc tures allowing outsiders to take actions that managers dislike in the event of poor firm performance, although a second-best solution, can be expected to do a more effective job of manipulating man agement incentives in productive directions.86 S3 See generally Erik Berglbf & Ernst-Ludwig von 1l1adden. Short- Te rm Ve rsus Lon::; Te rm interests: Capital Strucwre with Multiple in vestors, 109 Q.J. EcoN. 1055 ( 1994 ). S4 Paul Asquith et al., Anatomy of Financial Distress: An Examination of Junk-Bond fssuers. 109 Q.J. EcoN. 625 (1994), supplies some empirical conf·irmation for this point. 1l1 is study illustrates the different behavior of bank creditors and junk bond holders of distressed borrowers. The banks, unlike the junk bond holders. do not forgive principal (although they do extend maturity). The banks also tend to refuse refin�mcing outside of bankruptcy. ::->5 See Berglof & von Tbadden, supra note 83, at 1058-59. S6 See Mathias Dewatripont & Jean Tirole, A Theory of Debr and Equit,·: Diversirv of Securiries and Manager-Shareholder Congruence, 109 Q.J. EcoN. 1027. l 028 (1994 ). 436 CARDOZO LAW REVIEW [Vol. 19:409 Hart offers a more formal expression of this point. He notes that, given managers who derive no private benefits from control of assets, first-best results easily can be achieved (in a taxless world) with an all equity capital structure and a simple incentive compensation system. In a two-period situation he would simply make the managers' compensation depend entirely on the divi dend. That is, incentive compensation I should = n(dl + d2) , where TT is a small positive number. If the payment also covers liquidation proceeds-! = n[dl + (d2, L)]-the manager can be expected to make an optimal decision respecting liquidation at t = 1. If L > y2, the firm is liquidated at t = 1 and no indebtedness is needed to align management incentives.87 But managers do derive private benefitsfrom asset management, and in Hart's conception, the bribe TT required to align their incentives with those of the outside security holders is unfeasibly large. Accordingly, a com plex capital structure that includes control mandates must be interpolated. Now consider the steady payout convention respecting divi dends in light of the three factors identifiedas determinant in these models-management's powerful incentive to pursue private bene fits,the ongoing unobservability and unverifiabilityof management performance, and the remedial effect of a clear-cut, mandated pay ment to an outside investor. Add the fact that the accounting sys tem allows managers at least some room to manipulate the amount of reported earnings. To the extent of that allowance, amounts of both earnings and reinvested earnings are unverifiable. This ver ifiability problem suggests an information function for the steady dividend. Although the amount of earnings may be subject to question, the amount of the dividend, once declared, is not. This very concreteness lends the dividend a governance function related Similar observations have been made respecting the agency dynamics of investment within a firm. Arjit Mukherji & Nandu Nagarajan, Moral Hazard and Contractibility in In vestment Decisions, 26 J. EcoN. BEHJ\V. & 0RG. 413 (1995), models the situation or a principal investing in research and development projects. 1l1e authors show that if the principal receives verifiable"hard" signals concerning the quality of the projects during the development period, the principal will be able to make a full ex ante commitment to a project but that problems of opportunism and monitoring costs still will make for a second best result-the principal rationally will overinvest relative to the first best. In contrast. in a world holding out only "soft" noncontractible information prior to the last period. they predict unclerinvestment. K7 See HART, supra note 59, at 146-48. Note an interesting real world implication of these observations-incentive compensation should not come in the form of stock options but in the form of illiquid long-term positions in the stock. 1997] DIVIDENDS 437 to that of debt.88 Although it triggers no control transfer,89 if paid steadily it does perform an ongoing screening function.90 2. Contingent Control with Outside Debt and Outside Equity Two subsequent models, from Dewatripont and Tirole91 and from Hart,92 explore the interplay between two principal points emerging from the Aghion-Bolton contingent control model. First, the form and timing of rights attending the firm's payment stream determine incentives critical to the firm's value. Second, problems of noncontractibility limit the capacity of compensation schemes to move management incentives in productive directions. Both mod els include extensions to dividend policy. Dewatripont and Tirole present another two-period contin gent control model. Unlike the Aghion-Bolton model, this one in cludes outside debt and outside equity interests. Here, at t = 0, outside financing and incentive compensation arrangements are worked out and management chooses an effort level e. The level of 88 This is Fudenberg and Tirole's suggestion. See Fudenberg & Tirole, supra note 35. Their model addresses information asymmetries within a multidivisional corporation and looks to the dividend payment stream as a means of alleviating the problem. Here. the value of the dividend does not lie in a tendency to force management to step up the level of borrowing. but because the dividend is, like debt. a verifiable amount. As in other incom plete contracts models, noncontractible aspects of management employment generate the problem. Fudenberg and Tirole discuss the dividend as an incident to their study of the implica tions of the practice of income smoothing. Income smoothing is the manipulation of the timing of reports of revenues and expenses toward the end of making the earnings stream less variable over time. Since the practice can entail temporal manipulation of actual busi ness operations. it can increase the costs of doing business. It either results from exploita tion of flexibility built into the system of generally accepted accounting principles. or from direct manipulation of business operations. The latter case can include poor timing of sales, overtime paid to lead to acceleration of shipments, and disruption of the schedules of suppliers and customers. See id. at 76. In Fudenberg and Tirole's model, income smooth ing follows from a combination of risk aversion and private benetlts retention by divisional managers, encouraged by corporate headquarters' inability to commit itself credibly to long-term incentive contracts with them. The managers expect that poor performance will lead to central intervention that diminishes their private benefits. The managers have an incentive to boost earnings in bad times to forestall intervention. and to understate earn ings in good times as a way of saving for later. 1l1e model assumes that the costs of income smoothing are less than those of intensive auditing and monitoring that would be required in order to eliminate the practice. See id. at 80. t-:9 At least not directly. YO Sec Fudenberg & Ti role. supra note 35. at 88-90. But. on the other hand. since the dividend carries no promise to pay a set amount. it also lends itself to smoothing over time. See id. at 78. '! 1 See generally Dewatripont & Tirole. supro note 86. n Here. Hart's book. HART. supra note 59. draws on Oliver Hart & John Moore. A Theory of Debt Based on 1he lnalienobilily of Human Copiwl. 109 Q.J. EcoN. 84 1 (i994). 438 CA RDOZO LAW REVIEW [Vol. 19:409 e will be either high or low, with high e producing higher returns in later periods, but resulting in the incurrence of a utility cost U to the managers. At t = 1 the firm reports its firstperiod profit,np l, a verifiable amount that is determined by e, but which is not a suffi cient statistic for e. In addition, a signal s is realized at this point. The distribution of signal s also is determined by e, and s is a suffi cient statistic for the profit to be realized at t = 2, np2. But s is noncontractible and management compensation accordingly can not be made directly contingent on it. This model's distinguishing assumption is that the firm's capi tal structure accords decisionmaking power to either the outside debt holder or the outside equity holder at a critical moment. More specifically, immediately after t = 1, the outside holder ac corded this control power takes action a, which can either be acqui escence and continuance C in present management operations or stoppage S of management's continued pursuit of its business plan. Stoppage S can entail any number of subsequent actions, including liquidation, sale of a division or other downsizing, or redirection of investment policy. Whatever the action taken, for any given signal s, S entails less risky subsequent management than C, the probability distribution of which has fatter upper and lower tails. At t = 2, np2 is realized and income is shared in accordance with the contracts in the capital structure.93 Investment Realization of np I Realization of and contract and signal s np2 I I t=O t= l t=2 choice of f' action o taken The model examines two possible incentive compensation schemes for E, one constituted of private benefits only and the other including a salary. The purpose of any such scheme is of course to induce E to choose a high level e. But, given the modeL and in particular the noncontractibility of s, the optimal arrange ment must include a possibility of punishment in the form of a con trol transfer to outsiders who have the power to choose action 5. Since management always prefers C to S (whether or not C is effi cient), a structure that increases the possibility of such intervention as npl and s decline lends management an incentive to choose a high level of e, maximizing the possibility of a choice of C despite 3. Debt and Nl anagement Empire Building: Applications of Contingent Control Models to Dividend Policy Hart explores the same set of questions against a sequence of simplified two-period scenarios.% For Hart, as for Berglbf and von Thadden,97 short term debt imports the hardest discipline. To ex emplify this, he adjusts his two-period model of debt payment tim- 9-' See id. at 1035-39. If there is a trading market in the firm ·s securities these results should not change . or so Dewatripon t and Tirole argue. They note that if the market price perfectly ref lects s. it can be argued that as of 1 = 1 a stand-alone compensation scheme could be based on the stock price with optimal results. But they counter that so long as the managers derive private benefits from C. control rights will matter at r = 1. and that since the moral hazard problem persists to t= 2. compensation ought to rctl ect r = 2 results also. ln addition. the stock price might be distorted by noise trading or asset bubbles. See id. at 1039. Y:' See id. at 1046. '.ih As already noted. he j oins Dcwatripont and Tirole in asserting that management derives significant private benefits from controlling corporate assets and that this prefer ence is so pronounced that it cannot be controlled through an incentive compensation scheme. See supra text accompanying note 87. Y7 And for that matter. D ewatripont and Tirok. See Dewatripont & Tirole. supra note 86. at I 043-46. 440 CARDOZO LAW REVIEW [Vol. 19:409 ing (presented above )98 to remove the possibility of asset liquidation at t = 1. Theresult is that the faster payment schedule entailed in short term borrowing due at t = 1 becomes the ultimate showstopper against empire-building managers. Restating this point under conditions of certainty, so long as y2 > L, long term debt is optimal-no payments pl should be due at t = 1, and pay ments p2 should be due at t = 2. But if y2 < L then a large payment pl should be scheduled at t = 1. However, optimal results prove harder to obtain when uncertainty is interpolated. Hart models the situation where values will be either state a (yl"'y2a and La;, or state b (yl h y2b and Lb). He works though various possible interre lationships between levels of debt, states a and b, and first-best results. He findstha t for at least one scenario no first-best level of debt can be set-where y2a > La and y2b < Lh, and, further, yla + y2a 5, yl b + y2b but yl a 5, yl b· In state b liquidation should occur at t = 1, while in state a no payment pl should be due at t = 1. Setting the t = 1 payment pl at a level to trigger default in state b triggers a default and liquidation in state a where it is inefficient. Thus, in theory, high debt can trigger an inefficient liquidation just as low debt can prevent an efficient liquidation.99 Hart also examines the possibility that the insight bound up in this model might imply a set of terms for an all-equity capital struc ture. Since yl, y2 and L are observable but not verifiable, state contingent contracting is not feasible. Further, the manager's power-for-money tradeoff so favors power as to make it useless to offer it a direct payment to choose L at t = 1 in the appropriate case. But an appropriate constraint could follow from an all-equity capital structure that (a) forbade any additional equity financing after t = 0 and (b) provided for dismissal of the manager at t = 1 unless a large dividend payment dl * was paid. The provision for dl * substitutes for a high pl. Hart argues, however, that this eq uity substitution does not import the same flexibility as does the use of debt. With debt, payment mandates can be divided between t = 1 and t = 2, and thus the y2 flow can be taken into account as well as the y1 flow in the exercise of control. 100 Hart, having rejected the alternative of an all-equity capital structure, goes forward to consider the additional implications of his assertion that debt plays the critical role in solving the agency problems of equity. He adjusts his two-period model to consider lUI See id. at 136-38. TI1is model also is opened up for uncertainty. See id. at 138-39. 102 See id. at 140-4 !. 152-55. 442 CA RDOZO LAW REVIEW [Vol. 19:409 They observe that, given the parameters of their model, including its incentive payment component, there is no disincentive respect ing selection of a bad project. So long as the project increases the upper tail of the t = 2 distribution it increases managerial rent, and, interestingly, is not necessarily undertaken to the detriment of the equity interest. There can be no guarantee of separation (a man agement preference for no project ahead of selection of a bad pro ject) so long as the firm's financial ructurest is invariant upon project choice. The cure lies in assuring that capital structure does vary when new projects yielding higher rewards for management are undertaken.103 Concurrent short-term borrowing is the sug gested adjustment, bringing Dewatripont and Tirole into an exact alignment with Hart at the bottom line.104 4. Internal Finance Compared Another model, this one from Gertner, Scharfstein, and Stein, 105 complicates the foregoing analysis. Here, the incomplete contracts approach is applied to the incentives that prevail inside of firms with respect to projects financed with internally generated capital. Those incentives are then compared to those that follow upon short-term borrowing. More specifically, in this two-period model a stand alone single project firm financed by short-term bank borrowing (as in Hart and Moore) is played off against the same project undertaken as a division of a conglomerate corpora tion and financedthr ough cash flowsfrom corporate headquarters. The model shows that both internal and external finance have strengths and weaknesses. One strength of internal finance is a higher incentive to moni tor. Gertner, Scharfstein, and Stein definemonitoring as observa tion of a project that generates economically valuable ideas. Such ideas require implementation, and the owner of the project has the right to decide whether to implement. The model shows that inter nal finance and ownership in corporate headquarters generates more valuable monitoring than does bank financing. The bank under-monitors because it does not control the assets. If bank monitoring does give rise to an idea, the bank can of course hold up the borrower for a portion of the returns of implementation in 1 03 See Dewatripont & Tirole, supra note 86. at 1047-48. 1o-1 For a model that combines the thrust of this treatment with the insights of the signal ling literature. see Thomas H. Noe & Michael J. Rebello, Asymmetric Information, J'v/ano gerial Opportunism, Financing, il/1(1 Payout Policies. 51 J. FrN. 637 (1996). 1115 Sec Robert H. Gertner et al., Internal Ve rsus Ex rernol Capital i\1arkets. 109 Q.J. EcoN. 1211 (1994). 1997] DIVID ENDS 443 exchange for the idea's disclosure. But the bank still has the lesser incentive because it does not receive all the gains.106 A second strength of internal financestems from possibilities for internal as set redeployment. Here, Gertner, Scharfstein, and Stein make ref erence to the literature confirming that distressed assets tend to sell for distressed prices, 107 thereby relaxing this literature's usual assumption that upon default, the outside lender will realize L. In their model, L proves to be fully or nearly fully realizable only where the firm's assets are nonspecificones like land and buildings. As the assets become more firm specific, the bank's liquidation yield drops relative to the value of the asset. Internal finance emerges as superior to the extent that corporate headquarters can invest in multiple related projects, positioning itself to recoup more for redeployment upon failure.108 Outside bank financing also has advantages. The first con cerns manager incentives. Recall that the basic incomplete con tracts model of debt financing leaves the manager of the successful project in complete control after t = 1, retaining all value over p. This produces a high-powered incentive that cannot be built into an intra-corporate financing arrangement. Even if corporate head quarters could contract not to fire the successful divisional manag ers, it still owns the assets and thereby retains some power to reduce the managers' private benefits. With internal financing, then, gains are shared, incentives are less powerful, and effort levels are lower. 109 In addition, corporate headquarters is pre sumed to be in closer contact with the project managers than is the outside bank, implying a higher level of influence costs. This counterbalances the greater likelihood of productive monitoring with internal finance-headquarters is "more likely to be wined and dined and ultimately won over" by the divisional manager than is the bank. 110 E. Incomplete Contracts and the Dividend Puzzle 1. From Intrinsic Optimality to Noncontractibility The incomplete contract models bear a familial resemblance to earlier agency models of dividend policy. Here again, conven- 1 116 See id. at 1219-.20. 1117 See Asquith et aL supranote 84: Andrei Shleifer & Robert W. Vishny. Liquidarion Values and Debr Capacity: A i'vlarker Equilibrium Approaclz . 47 J. F1r-.:. 1343 (1992). I IlK See Gertner et a!.. supra note 105.at 1223-27. IO'l S' ee id. at 1220-23. 1111 /d. at 1229. 444 CA RDOZO LAW REVIEW [Vol. 19:409 tional dividend payout patterns are explained in terms of the prac tical likelihood that replacement investment dollars will be borrowed. Here again, the dividend forces self-serving managers to the capital markets, leverage enhances firmvalue , and manage ment will resist an optimal leverage policy. And here again, the control threat embodied in the possibility of default has a positive incentive effect. But there are also significant changes of emphasis. The origi nal agency models completed the incomplete corporate contract with a reference to trading markets in corporate securities. Infor mation was unproblematic because the stock price contained and communicated it. Multi-period models were unnecessary because the stock price was appropriately discounted for agency risks at the time of original issue. Thereafter, market forces kept agency problems in check on an ongoing concern basis. The Rozeff model and the Easterbrook model implied that the conventional dividend payout pattern is an optimal expression of market discipline. In a competitive equilibrium, after all, discretionary behavior cannot survive. 111 With the incomplete contracts models, market prices do not solve problems of information asymmetry. The solution to problems of observability and verifiability instead lies in the per formance of the debt contract. Since the steady dividend prompts borrowings governed by the debt contract, it emerges as a lesser version of the same solution to the problem of noncontractibility. These models also work from an underlying concept of agency re lationships quite distinct from that operative in the complete con tracts model of Jensen and Meckling and its many progeny in legal theory. Consider the assertion in Dewatripont and Tirole and Hart that an incentive contract cannot feasibly protect against empire building managers. Then compare the behavioral failings of Jensen and Meckling's managers-shirking and excessive consumption of perquisites. The latter conception of the agency problem invites easy disposition with cash payments and contractual prohibitions. Failing the appearance of such an effective contractual technology, the Jensen and Meckling model goes on to assume that capital and product market competition completes the incomplete contract, 112 fully disciplining culpable managers within a reasonable time . 1 1 I See Aaron S. Edlin & Joseph E. Stiglitz. Discouraging Rivals: Aianagenzenl Rent Seeking and Economic Inefficiencies, 85 AM. EcoN. REv. 1301. 1301 (1995 ) . 1 12 See S.A. Ravicl & E.F. Suclit. Power Seeking Managers, Profitable Dividends and Fi nancing Decisions. 25 J. EcoN. BEHAV. & ORe.. 241 (1994). 1997] DIVIDENDS 445 The empire builders in the incomplete contracts models make a different trade-off between money and the satisfaction of control ling assets, and are much less susceptible to market discipline. 1 13 Reconsider the Rozeff dividend model in light of this point. Like the incomplete contracts models, Rozeff's model accords a prominent role to the borrowed replacement dollar. But Rozeff also conceives of dividends as a substitute for incentive-aligning management stock ownership. The incomplete contracts models, although certainly assuming that some contracts align incentives better than others, perceive no chance for dividend payouts to fig ure into such an easily optimal alignment of incentives. The man agers, quite simply, draw too much satisfaction from the private benefits yielded by control of corporate assets. Removal from con trol is the only palliative. Reconsider also the Easterbrook dividend model. Easter brook located the benefit of the dividend in the monitoring that results when the firmhas to go outside for new capital. The incom plete contracts models, although quite sensitive to the importance of information and monitoring, shift emphasis to the form of the resultant financing. They relocate the point of agency cost reduc tion to the structural result that follows the reference to outside financing rather than in an investigatory process carried out by capital market actors. The shift makes sense. If management goes to the capital markets to sell additional equity rather than to bor row, resultant monitoring may be even more intense, but no gov ernance benefits will redound in future periods. However intense such capital market scrutiny, it can neither access the unobservable nor confirm (or falsify) the unverifiable. When monitoring does become the center of attention in in complete contracts models, particular incentive questions are asked about the design of the monitoring contract.114 These ques tions ask us to compare the different monitoring incentives follow ing from the different available modes of outside finance, toward the end of assuring not only that outsiders have information about 113 Sec id at 244. 114 See Ernst-Ludwig von Thaddcn. Long- Term Controcrs. 5'/wrr- Term Jnvestnzenr and ;'v!onitoring. 62 Rt:v. EcuN. Sn m. 557 ( 1995). Vo n Th adclcn takes up the problem created by management aversion to tl rst period inspection of a project by long-term outside linan cir.:rswith a power to terminate. Management may prefer a series or suboptimal short-term investment projects. His model takes up the alternative of a short-term credit tacility with an attached long-term commitment from a fi nancier who performs a "delegated monitor ing" function. See id. at 558-59: sec also Douglas W. Diamond. Financial Intermediation and Delegutc Monitoring. 51 REv. Ecot'. Snm. 393 (!984 ). 446 CA RDOZO LAW REVIEW [Vol. 19:409 i goings on inside the firm, but that the firmselec ts the best available I i investment projects. This literature requires that such questions be � asked because it no longer assumes that a reified "market" sees l through the veil that obscures outsiders' access to corporate infor mation, thereby automatically encouraging a choice of efficient projects. No such assumption is safe given asymmetric information and strategic behavior. 2. Noncontmctibility and Capital Structure Optimality The incomplete contracts models join the wider literature of information economics in abandoning the early agency models' complete reliance on market price and market discipline. As such, they leave the legal observer in an uncomfortable position of un certainty. This literature does not promise immediate evolution to a first-best competitive equilibrium in the absence of sovereign in tervention. It instead poses complex informational problems that are susceptible to solution only in stylized models. The models teach that debt solves problems of information asymmetry and suboptimal incentives by vesting a clear cut control transfer contin gency. Yet this yields no policy prescription because the vesting of the contingency turns out to implicate complex variables. In the incomplete contracts picture, debt makes production more efficient in those states of the world in which actions that enhance the value of the debt payment stream are also the actions that maximize the value of the firm. The holders of debt should step into control, and debt should be incurred in amounts and under terms so as to trigger control transfers on those states of the world. 115 Contrariwise, incorrectly set triggers reduce firm value. The models assert emphatically that there is such a thing as too much debt, and thereby identify it as a source of governance problems as well as governance solutions. This diagnosis is vari ously articulated. In the Dewatripont and Tirole model, too much debt means that conservative, variance-reducing business plans are chosen across too wide a class of situations. In the Hart and Moore model, excess debt means that liquidation occurs when continua tion realizes more value. In addition, Gertner, Scharfstein, and Stein show us that internal fmance possesses certain incentive and cost advantages relative to debt. In sum, leverage entails a trade off between a harder incentive structure that deters private benefit ' 1 5 Restating the point to rdlect the fact that the fulcrum provision goes to the size of the pavments clue at 1 = 1 and r = 2. the tirm should borrow an amount such that the lenders should assume control when it is optimal that they do so. 1997] DIVIDENDS 447 seeking and the risk that the control transfer that imports hardness will occasion suboptimal business decision making. Because these models work against a dynamic background that builds in uncer tainty about future events, they offer no calculus for the amount and terms of the debt contract that optimally balances the trade. The trade-off is further complicated by an additional factor. Our experience with the high leverage models of the 1980s has trained us to think in terms of a zone of alignment between high leverage capital structures and the interests of outside equity hold ers. TI1e Dewatripont and Tirole analysis inserts a cautionary note respecting this assumption when it suggests that there may be an expansive zone of alignment of the incentives of management and outside equity respecting levels of debt and internal financing. The reason is simple: Management and equity stand together in prefer ring that control not be transferred to the debt, and accordingly both react in a risk averse way respecting the level of debt in curred. It follows that if real world firms maintain suboptimally low levels of debt, then they do so because of an alignment be tween the incentives of management and equity.110 111i A question arises at this point in the analysis. In the 1980s models. the operative picture of the preferences of equity was derived by analogy from the Black-Scholes option pricing model. See Fisher Black & Myron Scholes. The Pricing of Oprions and Corpora1e Liabilities. 81 J. PoL. EcoN. 637 (1973). Under the derivation. equity was not risk averse and increased asset volatility enhanced equity values. How can that point be preserved in a description that simultaneously holds out a common element of risk aversion on the part of management and equity? The answer lies in the timing of the description. The element of risk aversion concerns the choice of the level of debt-the decision as to whether to borrow at all. In Black-Scholes terms. this becomes a choice among a menu of option contracts with different striking prices and durations. TI1e behavioral effects discovered by Black and Scholes follow only once one of the options on the menu is chosen. ll1e inten sity of any resulting propensity toward risk-taking will depend on that option's terms. TI1at selection will in turn depend on an ex ante appraisal. Given that this option hedges noth ing and stakes the entire equity investment. risk aversion can be expected to come into the selection process. Meanwhile. at the bottom line. the Black-Scholcs view of equity and the incomplete contracts literature make the common point that the maximization of the value of the equity payment stream and the maximization of the value of the firmproceed under materially different calculations. Finally. it should be noted that the identification of a common elt:mentof risk aversion common to management and equity respecting levels of inclebtc:dness does not imply an assertion of a complete iclentitv of interest between the two. Hart strikingly restates the core management-shareholder agency problem when he notes that. but for management's pursuit or the private benefits of asset control. an all equity capital structure compensating management with a small pro rata share of the divi dend payout pool would be optimal. The incomplete contracts models are onlv just beginning to articulate a description of management-shareholder relationships. At this stage. their descriptive contribution lies primarily in the negative point that optimal management incentives arc noncontractible and control transfer devices must be resorted to on a second-best basis. Presumably. other l 1l 448 CA RDOZO LAW REVIEW [Vol. 19:409 j j ! j 3. Noncontractibility and the Dividend I j 1 The dividend reemerges under incomplete contracts as a de ! i vice with two functions. First, it ameliorates problems of observa tion and verificationby transforming an unreliable book entry into a tangible payment. Second, it adjusts capital structures in the di rection of an optimal control transfer provision by prompting bor rowing. The dividend thus reemerges in a second-best world. Since the optimal level of debt is uncertain, the optimal level of retained earnings and dividends is also uncertain. The combination of that uncertainty, managements' pursuit of private benefits, and the shareholders' lack of control provides a working explanation of the dividend's real world tendency to fall short of an optimal amount. As a check, let us apply this working picture to Berkshire Hathaway. Can we use it to explain Warren Buffett's situation of low debt/17 no dividends at the parent level, substantial dividends from wholly-owned subsidiaries, and satisfied shareholders? We can. We see at once that the real anomaly here is that, empire building and high returns, for once, work in tandem. Given that, a low level of debt is rational-why take the risk if no additional performance incentive is necessary? For the same reason, steady dividends have no governance function to serve at Berkshire. They might serve an informational function, for Berkshire certainly presents a situation of information asymmetry. Special factors ameliorate the information problem in this case-first, Buffett's unusual reputation as an investor, and, second, his practice of writ ten reports to shareholders on the discretionary subj ect matter as to which managers and corporate reports usually are silent. 4. Noncontractibility and Legal Mandates The incomplete contracts models' distinction between noncon tractible and contractible corporate events and their identification of control transfer as the central mode of agent control also ad vance our understanding of an essential aspect of the structure of corporate law. Corporate law usually is enabling, but sometimes is mandatory, and contractibility appears to be implicated in the placement of the line between the two. Default rules that look to ward contractual solutions tend to apply to contractible subj ect matter. TI1e legal institutions of control transfer-the federal control transfer devices will be formulated. and the takeover and proxy contest will be brought into the incomplete contracts description in the future. 1 1 7 See LowENSTEIN. supra note l. at 272-73. 1997] DIVIDENDS 449 bankruptcy regime and the federal-state regimes for proxy con tests, mergers, and takeovers-present a contrast. They are thick with mandates, many of them contract invalidating. The dividend occupies an awkward place in this scheme. Since dividend and reinvestment policy determines the payment stream on common, it plays a significant role in shaping incentives that determine the operation of the corporation's mandatory control transfer institutions. Yet its unregulated status is fully consonant with a larger pattern, for debt incurrence is a business judgment matter also. Assuming that past institutional evolution assures the appearance of all feasible regulatory solutions to basic governance problems, it would seem to follow that no additional regulation of dividends feasibly can be interpolated because this subj ect matter is noncontractible. The robustness of this proposition is considered in this Article's third and finalpa rt. III. LEGAL STRATEGIES FOR PRODUCTIVE DIVIDEND AND REINVESTMENT DECISIONS: AN INCOMPLETE CoNTRACTS EvALUATION Do the incomplete contracts models, having offered a plausi ble description of familiar institutions, provide any insights as to how to improve them? At first inspection, they appear strongly to ratify present institutional arrangements. In corporate law, the div idend is the ultimate redoubt of management discretion, so long as management observes the formality of never publicly stating that the pursuit of private benefits really motivates its policy. 118 The standard explanation for this rule of discretion follows from the core insight of the incomplete contracts approach-that dividend and reinvestment decisions depend on decisions that follow from unobservable and unverifiable variables. Further inspection com plicates the analysis, however. These models supply a productivity explanation for observed dividend practice without taking the ad ditional step of asserting that the practice should be presumed to be optimal. TI1ey thereby implicate a theoretical possibility that an institutional reform, whether the result of contractual evolution or state intervention, could improve matters. They even signal a strat egy for the design of such a reform when they use the debt con tract's payment mandate to cure the problem of noncontractibility. As Hart points out, a mandatory dividend could have a similar ef.· 1 11' "I11 e classic case is Dodge v. Fo rd !Yforor Co .. 170 N.W. 668 (Mich. l919). 450 CA RDOZO LAW REVIEW [Vol. 19:409 fect.119 More generally, the institution of the dividend might be more productive if it took on some of the characteristics of debt. Threestrategies directed to the improvement of dividend and reinvestment policy appear in the legal literature: first, mandatory payout of earnings at the shareholders' option; second, dialogic in tervention by activist institutional investors or their agents; and third, stepped up disclosure mandates. Thediscussion that follows evaluates each of the three in light of learning from the incomplete contracts models.120 A. Mandatory Payout of Earnings at the Shareholders' Option Zohar Goshen/21 expanding on an earlier suggestion made by Merritt Fox, 122 recommends that public corporation shareholders be accorded a right to receive a pro rata share of each year's earn ings in cash, or, in the alternative, to receive a pro rata stock divi dend that in effect reinvests a pro rata share of earnings.123 Goshen offers a multi-step argument in support of this mandatory dividend option. Shareholders, he says, are unlikely to make suboptimal dividend and reinvestment decisions; therefore, a dividend option will cause capital to move in the direction of its best use. 124 The option would not cause material disruption to existing corporate financing practices, provided of course that man agement has been doing a good job of reinvesting earnings. Since the stock price already reflects the expected return on such a com pany's stock, each investor in the shareholder group presumably is satisfiedwith the rate of return; holders dissatisfiedwith the return already will have sold. So where legitimate growth prospects are on offer (or where no growth is on offer but a level of internally generated working capital is necessary to maintain cash flows), the shareholders can be expected to grasp the maximizing course and I I 9 See supra text accompanying note 87. 12o There is. of course. a fourth strategy-the takeover. This strategy is inevitably ac companied by a supplemental strategy for increasing the incidence of its employment in practice-the removal of state anti-takeover legislation. Since its advantages and disad vantages are exhaustively discussed elsewhere. it is omitted from the present discussion. 1 2 1 See Goshen. supra note 5. at 903-06. i 22 See MERRiTT B. Fox. FIN.-\NCE AND INDUSTRIAL PERFOR'VIANCE IN A DYNJ\i'vi iC ECONOMY 383-400 (1987). 1 23 See Goshen, supra note 5. at 906-09. 1 24 See id. at 905-06. Goshen points out that implementation of the option presupposes an overhaul of the tax system. Under present law. the retaining shareholder pays ordinary income tax on the stock dividend. See id. at 906- 17. 1997] DIVIDENDS 451 permit earnings retention.125 Of course, some firms reinvest suboptimally; as to these an increase in the amount of firmbo rrow ing can be expected. But, as Goshen points out, the bonding effect of a dividend option is less than that entailed in any such conse quent borrowing. This is because the payout mandate bound up in the dividend option is contingent on the existence of earnings, un like the absolute minimum performance bound up in the promise to pay.126 Goshen sees only one serious problem with a shareholder divi dend option. Coordination problems could cause shareholders to force a payout above the minimal reinvestment amount necessary to preserve the business.127 He makes a technical adjustment in order to solve this problem. Each shareholder who opts for reten tion does so conditioned on a minimum percentage of other share holders deciding to do the same thing; failure to meet the shareholder's stated threshold cancels the retention decision. Each shareholder thus will "reveal her true retention preference and avoid reinvesting in a firm with insufficientworking capital."128 1. Advantages Goshen makes no reference to incomplete contracts models of capital structure.129 Indeed, his operative assumptions approach those of a complete information model. Everyone grasps the cru cial information respecting value because the stock price accurately impounds all information. Furthermore, he recommends the divi dend option because it releases cash flows for reallocation by ac tors in secondary trading markets, not because it directs particular corporations toward an optimal incentive structure. Despite this, the dividend option has some appeal from an incomplete contracts perspective. 1 2s Some difference of opinion as to the prospects for the rate of return on reinvested sums will be inevitable. and not unhealthy. See id. 920-2 1. I 2o See id. at 904. 1 27 See id. at 92 1-25. 1 2s !d. at 924-25. I 2Y But he does of necessity reject the Modigliani-l'vliller hypothesis. If dividend levels. investment policies, and fi rm value were independent of one another. then dividend op tions presumably would make no difference. A management required to pay out all earn ings but which wanted to make a given investment would simply borrow the money and the end result vvould be the same as that following reinvestment. Recent empirical studies of dividend policy support the idea that dividends and invest ment policy arc indeed dependent on one another. See David A. Louton & Dale L. Domian , Dividends and lnvestmenL Further E111pirical Evidence. Q.J. Bus. & EcoN. . Spring 1995. at 53. 452 CARDOZO LAW REVIEW [Vol. 19:409 The advantage lies in the debt-like aspect that the option brings to the dividend payment stream. The step in the direction of debt lies in the creation of a constant possibility of a mandated payout: Depending on the shareholders' preferences respecting a payment stream, common stock has the potential to be an income bond without a maturity date. Although the option does not ac cord shareholders contingent control of the business (in a sense they already have it by virtue of their voting power), it does pro vide them contingent control of the portion of cash flows deter mined to be net earnings under Generally Accepted Accounting Principles ("GAAP"). Problems of noncontractibility are avoided through the combination of the reference over to GAAP and the vesting of absolute discretion in the shareholders. 1 30 The dividend option also has advantages when compared to high leverage as a solution to the problem suboptimal investment of free cash flows. As shown above, high leverage solves the prob lem absolutely subject to a long list of problems and costs. The dividend option builds in more flexibility, since it leaves the disci plinary decision to ongoing shareholder evaluation. Recall that Hart argues to the contrary, making a case for the superiority of debt over dividends. He bases that case, however, entirely in the context of a two period model. Debt, he says, imports discipline across the model's entire time frame, where a mandatory and large dividend only can operate at t = 1.131 The dividend option reveals that Hart misses a point. The corporate dividend and reinvestment problem arises on an open-ended time frame, and a dividend op tion, unlike a short- or long-term debt contract, spurs productivity across that indefinite succession of time periods.132 It also heeds the warning that borrowing carries intrinsic risks. Unlike a fixed payment, the dividend option can be relaxed during cyclical or other downturns. And in situations where discipline is needed, the option forces managers who wish to expand their empires to incur more debt. A dividend option also would materially alter management's incentives respecting disclosure of information about investment 131 l Management. of course. thereby acquires a costly incentive to manipulate GAAP to minimize reported earn ings and thereby protect cash t1 ows from share holder access. See supra note 88. But these costs arguably would be modesi . The existing accounting system delimi ts them. and the benefit of more productive investment practices would counterbal ance them. 131 See supra note 100 and accompanying text. l32 Of course a conso! bond. like a dividend option. could extend across an indefinite succession of periods. 1997] DIVIDENDS 453 policy. In the incomplete contracts framework, stock prices do not unilaterally (and heroically) solve problems of information asym metry, and management lacks incentives to disclose its knowledge about its investment policy. Given a dividend option, the zone of discretion to withhold information would shrink. Hypothesize a firm falling into Graham and Dodd's middle or low level growth category.133 Its shareholders would treat a management earnings retention request with some skepticism. If its managers wished to maintain or increase their stocks of retained equity capital, they presumably would be forced to mount a road show-taking their case out of the boardroom-and make credible presentations re specting anticipated project returns and costs of capital to the shareholders. Enduring information asymmetries thus would be ameliorated, if not cured. The talk would be cheap, but at least there would be talk. 2. Shortcomings Incomplete contracts theory lends the foregoing support to the dividend option with only one hand, however. It uses its other to withdraw the support. The withdrawal stems from informational problems. a. Information Asymmetries If this subject matter were contractible, the problem of subop timal reinvestment could be solved with a charter provision stating that the corporation's agents shall approve no investments as to which r < k. But, because the subject matter is noncontractible, such a provision would be unenforceable for all intents and pur poses. How would a shareholder plaintiff prove an alleged breach? Investment is a matter of valuation, and valuation is an intrinsically speculative exercise. The appraiser projects a stream of future cash flows,and then discounts them to present value by ascertaining and applying a required rate of return. Contemporary business prac tice makes available a range of methodologies for ordering such an inquiry, facilitating a degree of objectificationof the framework for analysis. The required rate of return, for example, can be hard ened by reference to numbers generated by comparable firms. But such hard numbers never come with a guarantee of accuracy-they are by definition generated in the past by different actors, and can only be applied based on the assumption that the future being pro- t33 See supra notes 17-lS and accornpanving text. 454 CA RDOZO LAW REVIEW [Vol. 19:409 jected will repeat that past experience. Some such assumptions are safer than others. And even in the safest case a relatively hard number is in the end only applied to a soft projection. In the real world of capital and investment, valuation ultimately depends on the judgment of the actor in control of the asset. When tested ex post by a legal decision maker, such a valua tion judgment at best can only be measured against a range of plausible results, and then only at considerable cost. There is no litmus test. This means that even if particulars respecting manage ment investment decisions were observable-and they are not under prevailing disclosure practices-it would not follow that ·i suboptimal choices could be proved reliably. The bad faith actor is j difficult to detect; even a long sequence of suboptimal returns could be put down to overly optimistic projections or subsequent bad luck, rather than to consciously suboptimal reinvestment poli cies. Corporate law thus has good reasons for remitting these deci sions to the realm of discretion. Where judicial scrutiny of financial policy does occur, as with certain mergers,1 34 the impossi bility of verification causesreview to be process-based. That is, the reviewing authority does a circumstantial check only, looking to see whether the deciding corporate agent appeared to be doing a conscientious, disinterested job. As noted, the dividend option obviates the verification prob lem by giving each shareholder absolute decisional power. The problem is that shareholders thus empowered still cannot observe the firm's set of new investment opportunities. A management group, once pressed, can attempt to ameliorate this asymmetry through disclosure. But, as also noted, that talk still would be cheap. The disclosing managers would be interested in the out come, and verification would be expensive and unreliable. Share holder choices respecting dividend options would not differ structurally from the choices shareholders already face when they buy and sell stocks. How then can these outside investors be ex pected to do the best possible job of ascertaining the quantum of r > k investments available to the firm? Goshen asserts that they can, assuming that the shareholders' attention will be focused by their staked capital and the possibility of gain and loss. The managers, in contrast, suffer the disadvantage U-+ See, e.g, . Cinerama. Inc. v. Te chn icolor. Inc. . 663 A.2d l l 56 (Del. 1995 ) (discussing the duty of care of directors of a linn acquired in friendly merger): We inberger v. UOP. Inc . . 457 A.2d 701 (Del. l983) (discussing the duty of loyalty of directors of subsidiary merging into parent). 1997] DIVIDENDS 455 of playing for private benefit. Against this assertion we can bring to bear every factor pointed to in the literature of stock market pricing imperfections. In the noise-trading counter story,135 many stockholders can be expected to look backwards to the immediate past and chase trends. Perverse projections follow immediately. Given widespread trend chasing and a dividend option, the high growth firm ofthe recent past would retain all earnings, while the middle or low level firm would pay everything out. A group of decisionmakers thus biased could not possibly make first-best decisions.136 But might such shareholders still do a better job than manag ers biased by pursuit of private benefits? Perhaps, if the only pri vate benefit pursued by the managers was the empire builder's satisfaction of asset control. But today's management incentive picture appears to be somewhat more complex. Managers also de rive private benefits from reputations for effective performance. Where twenty years ago such a reputation depended on company growth, evaluations today are keyed to shareholder value. Reputa tional incentives accordingly tend to encourage more care in the choice of new investment opportunities. The problem of subop timal reinvestment has not gone away, but it is hard to predict that an under-informed shareholder with a pure financial incentive will make reinvestment decisions superior to those of a better-informed manager with a complex of motives. One suspects that results would vary from company to company-some deadwood manage ment teams would get a dose of needed discipline, and other con scientious and talented teams would be forced to borrow more than an optimal amount or simply to pass on good opportunities, the value of which proved difficult to communicate credibly.137 135 See Andrei Shleifer & Lawrence H. Summers, Th e Noise Tra der Approach w Fi nance, J. EcoN. PERSP .. Spring 1990, at 19, 19-22. t 3o For a recent empirical study supporting this point, see Josef Lakonishok et al.. Con !rarian !nvesrmenr, Exrrapolation, and Risk, 49 J. FrN. 1541 (1994) (arguing that statistical analysis of stock prices and returns over time suggesting that investors in ·'glamour '· �tocks suffer from a cognitive limitation). t37 We could make Goshen's mandate more situation-sensitive by making its presence contingent on a statistical signal that distinguishes firms with extraordinary investment op portunity sets from !inns with suboptimal sets. That is. all public firms would start life without the dividend option. which would be triggered by the occurrence of the signal-in effect at r = I. Any number of concrete events suggest themselves as appropriate identill ers of this sort of corporate maturity-two or three years of earnings growth below a level fi xed by reference to current market conditions, or perhaps a stock price stalled or falling in relation to some larger index. Any signal chosen would carry some imperfections. and also be in danger of obsolescence given changing conditions. But a rough separation of sheep and goats presumably could be engineered. In the encl. however. this modification 456 CA RDOZO LAW REVIEW [Vol. 19:409 b. Perverse Incentives The dividend option promises to take us closer to a first-best corporate investment practice by removing the dividend and rein vestment decision from the influence of private benefitpursuit by transferring it to a set of decision makers possessing purely finan cial incentives.138 In making this promise, Goshen follows con tractarian theory to assume the aggregate of shareholders provides a perfect proxy for the maximizing firm. The incomplete contracts approach, in contrast, depicts each class of outside interest holder in a distinct governance incentive posture that follows from the terms of the holders' payment stream. Given such a picture, ques tions arise respecting the incentives of under-informed, option holding shareholders. To access these questions, let us look once again at Warren Buffett. He stands out among American investment managers not just as a bargain-hunter, but as a bargain-hunter patient enough to wait for the business cycle to turn before buying. The herd, in con trast, is famous for pulling up stakes when the equity averages go south and reinvesting in debt securities to weather the cyclical shock. How, given such a behavior pattern, would a dividend op tion work? One possibility is that investors who chose reinvest ment in good times would change their options in bad times, taking advantage of the debt-like characteristic of their holdings. Although they would not bootstrap themselves into bankruptcy priority in extremely bad times, they still would see an increase in periodic income that partially would cover cyclical stock price reverses. An institutional holder, such as a bank or insurance com pany, experiencing a cyclical round of defaults on debt invest ments, would have a similar incentive. The managers of the company thus optioned would be grappling with a recession and the consequent tightening of lenders' credit standards. They might be squeezed indeed. Such a squeeze, taking a toll on the stock price, would not make a switch from earnings retention to a cash option irrational from the point of view of an individual share holder. Unlike a manager, this actor has an ever-present exit op tion. Playing chicken and testing the patience of the rest of the shareholder group might very well make sense; if the game fails on the first round, the stock can be sold. reduces the problem of uninformed shareholder choice only by reducing the number and type of firmssu bj ect to the option. The substance of the problem remains. t 3X Given perfect information. only managers with perfect incentive compensation con tracts could do as good a job. 1997] DIVIDENDS 457 Product market competition also could give rise to unexpected problems. Bolton and Scharfstein have a model showing that an endogenous financial constraint imposed to assure that a firm does not divert resources to itself, in their case short-term borrowing, can be costly in a competitive environment. The financial con straint gives the firm's rivals a predatory incentive. They step up the price competition, and thereby increase the chance that the firm's lenders cut off funding, inducing its premature exit.13" Pay out of all or most corporate earnings, taken alone, would be un likely to lead to this dire consequence. But it could make a significant causal contribution. For example, if the shareholders opted to take an amount greater than that needed to sustain the business, the firm's debt level would become suboptimally high with resulting vulnerability in the product market. Even if the debt was not excessive in theory, predatory conduct by other firmscould make it excessive in fact. Alternatively, management groups facing a world in which equity financingimplies an unstable claim on firm cash flows might resolve doubts in favor of higher proportions of debt financing, thereby suboptimally exposing themselves to dis tress in the event of product market reverses. Finally, a dividend option could be a vehicle for shareholder self-dealing in some situations. Large block share he· Jers particu larly might have incentives to opt for cash, thereby se ·, Jing a signal to other shareholders to do the same thing. They might, in so do ing, soften up the firm and its stock price for a later takeover or partial combination. 140 B. Noncontractibility and Intervention by Institutional Investors The mandatory shareholder vote for the board of directors is corporate law's primary response to the problem of noncontrac tibility. The vote is a contingent control transfer device. Since in cumbents may be terminated for any or no reason, no problems of observability or verifiability are implicated. Of course, control is transferred to a group of replacement agents rather than to an other class of security holder, as occurs upon default on borrowing. But, presumably, the replacement agents get control on an under taking to step up the payment stream on the equity. The real prob- 1 39 See Patrick Bolton & David S. Scharfstein, A Th eory of Predation Based on Agency Proble111s in Financial Contracting, 80 AM. EcoN. REv. 93 (1990). 1 -1! 1 See Jeffrey N. Gordon. Shareholder lnirimive and Delegation: A Social Choice and Came Th e01·eric Approach ro Corporare Law, 60 U. CrN. L. REv. 347, 376-81 (1991). 458 CA RDOZO LAW REVIEW [Vol. 19:409 lem with the vote lies in the inhibiting effect of economic and legal barriers to collective action by shareholders.141 Imagine a world in which shareholders used their votes to elect activist board representatives, and in which managers as a re sult lived in constant fear of replacement. Dividend and reinvest ment policy presumably would be a primary focus for such directors. Problems of observability would still surface. Board room presence, taken alone, does not give an outside director facts with which to expose a suboptimal investment project, for example. But, armed with even a modicum of technical expertise, such direc tors could be expected to subject the credibility of management's investment program to plausible inspection. Such intense monitor ing, together with a politics of corporate tenure, could materially lessen the intensity of the dividend and reinvestment problem, avoiding some of the perverse effects of the blunt-edged control transfer devices modeled in the incomplete contracts literature. 1. Th e Institutional Shareholder Movement The initiation of such a monitoring regime is the core objective of the institutional shareholder movement. Its strategists have of fered plans for cost effective shareholder-initiated monitoring of noncontractible corporate affairs and negotiation (contractual and otherwise) of reductions of the costs of management influence142 within the firm. Given certain legal adjustments, the strategists have said, prospects for financial gain by themselves will induce governance initiatives by institutional investors.143 In the alterna- 1 4 1 Historically. shareholders of public companies are an Olsonian latent group. See Edward B. Rock. Th e Logic and (Uncertain) Significance of Jnstitwional Shareholder Ac tivlsnz. 79 GEo. L.J. 445, 455-59 (1991). In other words, a collective good-active monitor ing of management-would make them better off given proportionate distribution of its costs. but the law provides no cost sharing mechanism, and the free rider problem prevents the emergence of a volunteer or group of volunteers with an incentive to provide the good. Given dispersed shareholdings, the nontrivial costs of active monitoring, and the alterna tive of exit through sale, the benefits obtainable without investment in monitoring exceed the benefits obtainable from investment. See id. at 455-56. In addition, rational apathy can prevail when the system mandates that matters be presented for shareholder approval. The rational small shareholder does not invest in information respecting governance mat ters. given the likelihood that the collective action problem prevents an effective group response. See Joseph A. Grundfest, Jusr Vo te No: A Minimalist Straregy fo r Dealing wirh Barburiuns !mide rhe Gates. 45 STAN. L. REv. 857. 910 (1993). As Edlin & Stiglitz, supra note lll. at 1301. comment, the puzzle for solution here is the existence of any manage ment disciplinary effect at all. t-->2 See Paul Milgram & John Roberts. An Economic Approach to !nfluence Activities in Orguni::.arions. 94 AM. J. Soc. Sl54, S156 (Supp. 1988). 1 43 Collective action theory allows for the possibility that a subgroup of a latent group will organize and provide for the public good if the benefits from action to each member of 1997] DIVIDENDS 459 tive, institutional votes could be used to nominate and elect expert outside monitors, 144 and the increased incidence of the placement of substantial blocks of shares with Buffett-like institutional owners.145 Unfortunately, no volunteers have appeared to make the fi nancial investments necessary for real world testing of these ambi tious proposals.146 Instead, institutional initiatives against badly managed firms have taken the form of discrete, issue-based voting contests that focus on short term results.147 Such exercises have low out-of-pocket costs and appear to be driven by the selective incentive of reputation rather than by a pure financial incentive. The leadership role has been taken by a narrow segment of institu tional agents-public pension fund managers whose indexed port folios reduce their share of immediate financial returns, but whose independence from management influence creates a possibility for reputational enhancement through constructive anti-managerial the subgroup exceed the costs incurred. See RussELL HARDIN, CoLLECTIVE AcTION 41 (1982). Increased concentration of shareholdings in institutional hands makes it conceiva ble that investment in monitoring might be cost beneficial for institutional subgroups. See Bernard S. Black, Shareholder Passivity Reexamined, 89 MrcH. L. REv. 520. 525 ( 1990). Concentration also promises to mitigate the rational apathy problem. The decision whether to become informed about the governance issue depends on the costs and ex pected benefits of the effort and the initiative's probabilities of success. 1l1e cost is in dependent of the number of shares held. With individual shareholders holding larger proportionate stakes, the expected returns from a given information investment go up, as does the proponent's probability of success. See id. at 585-89. Subgroup formation depends on the size of the group. the cost of action. and the magnitude of the benefit to be obtained. Proponents of law reform to facilitate share holder participation direct most of their attention to the first two factors. Since the number of members needed to form a subgroup declines as ownership concentration goes up, the proponents argue for relaxation of regulatory barriers that impede the accumula tion of large holdings in given firms by single investors or organized groups of investors. See id. at 579-80. The proponents also circulate blueprints for cheap strategies. since . as the costs of a given initiative go down, subgroup formation can go forward with a lower level of concentration and a lower projected probability of success. See Grundfcst. supra note 141. at 927 (explaining minimum cost strategy of "just vote no" campaigns). l44 See Ronald J. Gilson & Reinier Kraakman, Reinvellling the Outside Direcror: An Agenda fo r !nstiwtional Investors. 43 STAN. L. REv. 863 (1991 ) [hereinafter Gilson & Kraakman, Th e Owside Director]; Jeffrey N. Gordon, Jnsritwions as Re/arionaf fri vl!s!ors: A New Look at Cumulative Vo ring, 94 CoLUi\-1. L. REv. 124. 133-42 (1994). 1 45 See Ian Ayres & Peter Cramton, Relarional ln vesring and Agency Theory. L'i C;R oozo L. REv. 1033 (1994); Ronald J. Gilson & Reinier Kraakman. fnvesrmcnr Co111punies As Guardian Shareholders: The Place of the iv!SJC in rhe Cotporate Governance De bare. 45 STAN. L. REV. 985 (1993). 1 46 See William W. Bratton & Joseph A. i'vl cCahery. Regularory Competition. Regularorr Capture. and Corporate Se!f-Regulution. 73 N.C. L. R�o.v. 1861. 1904-06 (1995 ). 147 See Robert C. Pozen. Institutional Investors: Th e Relucranr Activists. H,c\RV. Bus. REv., Jan.-Feb. 1994, at 141, 145-46. 460 CA RDOZO LAW REVIEW [Vol. 19:409 political activity.148 They have had some successes. Their commu nicative courses of action have prompted preemptive negotiations and concessions by managers, and, in some cases the termination of the chief executive by the outside directors.149 These shareholder threats are credible because they impact on the reputational interests of chief executives and independent board members. A campaign, by its very existence, declares that the target executives possess undesirable characteristics, 150 de tracting from their standing in the business community,151 and in some cases, their marketability. Extraordinary risk aversion to such reputational impairment can be expected on the managers' J48 These events confirm the predictions made by the theoretical counter story. Even given legal adjustments. governance initiatives realizing the full promise of cooperative gain through enforced self-regulation cannot be expected. Tw o points are emphasized. First, agency relationships within investment institutions create disincentives that prevent subgroup formation. even assuming a projection of a positive return to the subgroup from an investment in governance. Since the individual manager's performance is measured against the performance of the market as a whole and subgroup investment benefits the market as a whole, successful governance investments do not necessarily improve the indi vidual manager's 1=-erformance profile. See Rock, supra note 141, at 473-74. Second. the benefits of cost-intensive relational investment remain underspecified. In theory. these lie in informational access and ongoing constructive criticism by the institutional monitor. In practice, underperforming companies are publicly identified in the ordinary course, and standard remedies respecting investment policies, incentive schemes, and governance structures are part of the conventional wisdom. To the extent that institutions cheaply can tie the communication of these points to credible threats against target managers, the avail able set of governance benefits can be secured through discrete engagement. Incentives for more substantial investments in ongoing relationships remain speculative. absent a spe cial technical capability on the part of the particular monitor. As a result, risks of perverse incentives and commitment problems come to the fore of the relational picture. A strate gically placed institutional holder could opt for side payments from management in prefer ence to public-regarding informational development, or, given a hostile tender offer, the institutions in the subgroup could defect from an implicit undertaking to management to be patient. See generally Ayres & Cramton, supra note 145; Edward B. Rock, Controlling the Dark Side of Relational Investing, 15 CARDOZO L REv. 987 (1994). 149 The means of access is the precatory shareholder proposal, a medium for nonbind ing, shareholder-initiated voting proposals made available by preemptive mandate under federal proxy rule 14a-8. Institutions began making these proposals in the late 1980s in reaction to expanding legal constraints on takeovers. The first generation of proposals concerned poison pills. but in subsequent years the subject broadened to cover the share holder voting process. and process and structure rules designed to make boards more effec tive in monitoring and designing incentive arrangements. See Gilson & Kraakman. Th e Outside Director, supra note 144, at 868; see also Grundfest. supra note 141. at 931. In the alternative, the proponent announces performance dissatisfaction directly and invites others to concur by voting no on management proposals. See Grund fest. supra note 141. at 931 . 1 50 See Grundfest. supra note 141. at 927-28. 151 Cf James G. March & Zur Shapira. Managerial Persp euives on Risk and Risk Ta k ing, 33 MG1\tT. SCI. 1404. 1413 (1987) (stating that managers are concerned about their reputations for risk-taking). 1997] DIVID ENDS 461 part, if, as seems reasonable, we can assume that their employment contracts, like investment contracts, are incomplete. With execu tives, incompleteness means that the contract does not fully com pensate for tenure insecurity and the costs of changing jobs. 152 More broadly, the appearance of a vocal shareholder interest group changes the manager's institutional environment. The insti tutions articulate a normative challenge to the manager's conduct of the business. Their challenge has a more destabilizing effect than ordinary external criticism, due to their equity investments, long-term presence, and ability to marshal votes respecting both present and future matters for shareholder action. They represent an unstable sector in the larger domain of institutional relation ships with which the manager deals. By negotiating, the risk averse manager seeks to stabilize, and, hopefully, influence the relationship.1 53 2. Dividends and Institutional In vestors Tw o well-publicized exceptions to the general rule that the in vestor activists fighttheir battles in the form of discrete issue-based voting contests should be noted here. These are the Carl Icahn Bennett LeBow proxy fight at RJR Nabisco and Kirk Kerkorian's campaign to gain representation on the Chrysler board. Each bat tle concerned dividend policy, and each was led by a large block holder pursuing an extraordinary short-term return on a common stock investment. More particularly, the Icahn-LeBow proxy con test was a step in a larger campaign to pressure RJR to separate its food and tobacco businesses by means of a spin-off. Kerkorian, in contrast, sought an extraordinary cash dividend, targeting a huge pot of cash-amounting to $7.5 billion-accumulated by Chrysler as a reserve to financethe retooling of its production platform dur ing the automobile industry's next cyclical downturn. Icahn and LeBow lost their proxy context;154 Kerkorian and Chrysler settled, 1 2 5 See Milgram & Roberts, supra note 142, at Sl58-59. 153 See March & Shapira. supra note 151. at 1414. 1 54 The sequence of events was as follows. LeBow received FTC approval to purchase up to 15% of RJR's stock on the open market in August 1995. In early October 1995. LeBow announced a partnership with Carl Icahn to purchase a 4.8% stake in RJR. LeBow then waged and won a nonbinding consent solicitation of RJR shareholders to compel RJ R to spin off the Nabisco food unit in February 1996. RJR countered to appease restless institutional investors by increasing its dividend 23% and setting a stock buy back objec tive. LeBow then made what is taken to be a tactical error. He announced a proposed settlement of several tobacco liability suits pending against his own tobacco company. Lig gett. That settlement caused tobacco stocks to decline across-the-board. LeBow conceded 462 CARDOZO LAW REVIEW [Vol. 19:409 with Kerkorian getting a representative on the board and a divi dend increase, but with management retaining its war chest.155 The present interest of these cases lies in the absence in each of serious problems of observation and verification,despite a focus on a dividend proposal. Each case was easily stated, without a need to review a complex valuation process. In addition, each of RJR's food and tobacco combination and Chrysler's war chest was in plain sight. RJR presents the more typical case of the two. Spin-offs and other forms of corporate unbundling are, along with CEO termination, primary items on the institutional investor agenda.156 Indeed, the frequent occurrence of corporate un bundling through spin-offs in recent years157 demonstrates both the rise in shareholder influence and the concomitant shift in the reputational influences on managers. Contrariwise, the spin-off's very prominence on observed in stitutional investor agendas demonstrates the restrictive effect of information asymmetries. Direct assaults on periodic retention de- defeat in the RJR proxy contest one month later. See Suein L. Hwang & Steven Lipin. LeBow Quits Fight to Win RJR Control, WALL ST. J .. Apr. 17. 1996. at A3. The fight is not yet over. As of this writing. lcahn. having dissolved his partnership with LeBow, has renewed the fight for a spin-off of the Nabisco food group. conducting meetings with RJR management and threatening to call a special shareholders' meeting. See Icahn Is Try ing Again to Fo rce RJ R to Split Fo od, To bacco Units. WALL ST. J.. Aug. 23. 1996, at B3. 155 Kerkorian tried several tactics to accomplish his goal of a payout of the war chest over a two year period. First, he launched and aborted a $20 billion hostile takeover bid. See Robert L. Simison et al., Putting Chrysler in Play. WALL ST. J.. Apr. 14. 1995. at A4. Next, he raised his stake in Chrysler from a 9.2% stockholding to a 13.6% holding through a $700 million tender offer and threatened a proxy fight. See Gabriella Stern et al.. Chrysler Fa ces New Pressure fr om Kerkorian, WALL ST. J .. June 27. 1995, at A3. In the end he settled for a five-year standstill agreement with the board. See Angelo B. Henderson & Gabriella Stern, Chrysler Corp., Ke rkorian Reach a Five- Ye ar Tr uce Agreement, WALL ST. J., Feb. 9, 1996, at A3. Chrysler. attempting to mollify Kerkorian. raised dividends five times over the two year period. It also announced a $2 billion stock buy-back and a two for-one stock split. The standstill agreement requires Kerkorian to keep his stockholdings below 13.7% and prohibits him from launching any proxy fights or takeover attempts. In exchange. Chrysler a'greed to the placement of a nominee on Chrysler's board. the announcement of an additional share repurchase program. and a change in some corporate governance policies. 1 56 Sometimes the two are combined: a CEO target with a weakening internal political base resecures his position by announcing a spin-off of a substantial segment of the corpo rate group. See, for example, Richard Gibson. Quaker Oats Fe eling Pressure fo r Big Changes in the Wa ke of rhe Fizzled Snapple Acquisition . WALL. ST . .1 .. July 25. 1996. at C2. recommending a spin off of Quaker Oats beverage business: ··Some investors are betting that Quaker Oats Chairman Bill Smithburg must act soon to bolster Quaker's stock price if he doesn't want to become Cap'n Crunched.'' 157 See Steven Lipin & Randall Smith. Spinoff's Flourisli. Fueled Bv Ta x Status. fn vesror Pressure, and Stock Performance, WALL. ST. L June 1.5. 199.5. at Cl. 1997] DIVIDENDS 463 cisions are uncommon-Kerkorian's attack on Chrysler's war chest is the only well-publicized case. A spin-off provides an indirect route to a similar end, one that surmounts the information prob lem. To see this point, we can turn once again to the writings of Warren Buffett. Suboptimal dividend and earnings retention deci sions, he says, are particularly likely where the managers of an out standing business add other businesses to the corporate entity: Many corporations that consistently show good returns both on equity and on overall incremental capital have ... em ployed a large portion of their retained earnings on an economi cally unattractive, even disastrous, basis. Their marvelous core businesses ...camouflage repeated failures in capital allocation elsewhere . . . . The managers at fault periodically report on the lesson they have learned from the latest disappointment. They then usually seek out future lessons.158 To push for a spin-off, then, is to push for a corporate structure that presents a lesser likelihood of suboptimal earnings retention. An immediate payoff in the form of an increased stock price also can be expected. 3. Law Reform If it became the practice for institutions to nominate and elect their own board representatives, direct challenges to questionable dividend and reinvestment policies presumably would occur more often. Once on the board, the representative at least would have a chance to access the relevant information. Any law reform insti tuted as a means to promote institutional board representation 159 thus also promotes more productive dividend and retention decisions. Pending realization of the full program of relational engage ment between institutions and managers, we can expect the institu tions to continue the practice of discrete, issue-based engagement of managers of poorly performing companies. Tw o alterations in the regulatory landscape-one at the federal level and the other at the state level-could facilitate increases in the scope and intensity of these piecemeal engagements. At the federal level, dividend policy could be admitted to the list of subject matter suitable for shareholder proposals.160 At the state level, standing barriers to I :'iS Buffe u Essuvs . supra note 9. at 126: see also LowENSTEIN, supra note 34, at 124-25. !59 For a list of regulations singkJ out for amendment or removal, sec Mark J. Roe, A Polirical Th eory of A111ericun Corporore Finance. 91 CoLUi'vi. L. REv. 10. 26-27 (1991 ) . too Under Rule 14(a)-8(c)( L3). specific amounts of cash or stock dividends are inappro priate subject matter for shareholder proposals. ll1 e SEC. pursuant to this rule, has con- 464 CA RDOZO LAW REVIEW [Vol. 19:409 shareholder access to the corporate charter could be removed.161 Both reforms would increase chances for access to the corporate contract for the inclusion of shareholder-initiated terms. Of course, in so doing, neither could bring about a quick solution to the dividend and reinvestment problem. Since each would be di rected to the corporate contract, each would suffer the limitations of the existing menu of contracting techniques. The menu contains a list of second-best solutions and partial palliatives-most promi nently, shareholder payout options,162 set payout percentages, and disclosure rules. Despite these limitations, it would be instructive to see whether any of these provisions garnered investor support or otherwise proved useful to proponents in company-specific con texts. And it always remains possible that someone might invent an effective regulatory technology in the future. Were that to hap pen, corporate law's process rules should provide a ready basis for its imposition at the shareholders' behest. Even given a first-best contractual technology, voluntary management adoption cannot safely be predicted. C. Mandatory Disclosure Warren Buffett comments as follows on management commu nications respecting dividend and reinvestment decisions: Dividend policy is often reported to shareholders, but sel dom explained. A company will say something like, "Our goal is to pay out 40% to 50% of earnings and to increase dividends at a rate at least equal to the rise in the CPl." And that's it-no analysis will be supplied as to why that particular policy is best for the owners of the business. Yet allocation of capital is cru cial to business and investment management. 163 Indeed it is. Even now, after six decades of experience with a mandatory disclosure system and a decade of experience with shareholder activism, meaningful disclosure practices respecting dividend and reinvestment decisions have yet to evolve.1M tinually issued no-action letters to registrants seeking to exclude shareholder proposals regarding the declaration of dividends. See. e.g.. Shop Te levision Network, Inc.. SEC No Action Letter. 1991 WL 176906 (S.EC.) (May 20. 1991). IAI Or. to the extent state level reform is impossible because of management capture of the state corporate lawmaking institutions. the federal government should intervene to mandate such access. See Bratton et a!.. supra note 43. at 1936-47. I A2 See supra text accompanying notes 121-28. 163 Buffett Essays. supra note 9. at l23. 164 SEC disclosure requirements pertaining to the pavment and declaration of dividends are very general. Regulation S-X requires the registrant to disclose the amount of the dividends for each class of share. See 17 C.F.R. � 2!0.3-04 ( 1994). The Regulation further 1997] DIVIDENDS 465 The noncontractibility of the subject matter explains this in part. In the rare case where a manager makes a disclosure about a dividend or reinvestment decision, the statement amounts to cheap talk because the decisions' bases are unobservable.165 Smart money looks for signals with more credibility. The absence of in vestor pressure for a more forthcoming practice accordingly makes sense. For the rest of the explanation we can look to the SEC's early tradition of limiting the disclosure mandate to hard, verifiable information, 166 and its more recent problems in constructing a via ble safe harbor for projections disclosure. 167 Reinvestment deci sions are made to finance investments, and investments follow from projections. Mandated disclosure of particulars respecting re investment decisions thus sounds suspiciously like a mandate to disclose internal projections. Theeffect of this rule of nondisclosure-once coupled with the state law allocation of utmost business judgment protection-is to embed dividend and reinvestment decisions in the deepest and darkest corner of the corporate black box. We lack any sense of the quality of boardroom practice respecting dividend and rein vestment decisions, in contrast to our extensive, litigation-gener ated knowledge of practice respecting mergers and takeovers. 16:::; A long list of questions can be asked. Do boards in fact receive peri- requires that any restrictions on dividend payments be noted in the notes accompanying financial statements. See 17 C.F.R. § 210.4-08( e )(1) (1996). No other significant informa tion respecting dividends needs to be reported in periodic financials filed with the SE C . See generally Regulation S-X, 17 C.F.R. § 210. 1-01 to 210.6-07. Regulation S-K requires the registrant to disclose cash dividends declared per common share in its selected financial data, and permits the disclosure of any other additional items that the registrant believes would enhance an understanding of other trends in its financial condition. See 17 C.F.R. § 229.301 instr. 2. 165 1l10se skeptical about signalling explanations of dividend payout patterns dismiss dividend increases themselves as cheap talk. See Harry DeAngelo et al., Reversal of Fo r fllne: Dividend Signaling and the Disappearance of Sustained Earnings Growth, 40 J. FtN. EcoN. 341 (1996) (supporting the conclusion that dividend increases are not systematically reliable as signals). 1 o6 For a description of the SEC's historical policy, see George J. Benston, The Ejfecrive ness and Effe cts of rhe SEC's Accounting Disclosure Requireme111s. in EcoNOMIC Poucy AND THE REGULATION OF CoRPORATE SEcURITIES 26-30 (Henry G. Manne ed., 1969). Ih7 Necessitating congressional intervention. See Private Securities Litigation Reform Act of 1995, Pub. L. No. 104-67. § 102. 109 Stat. 737. 749 (codified ns amended at 15 U.S.C. § 77z-2 (Supp. I 1996). inserting new section 27A into the Securities Act of 1933 with safe harbor for forward-looking statements). IGK There. the application of fiduciary rules of care and loyalty has resulted in the evolu tion of a process standard. Boards are not required to bring back some minimum price for the company for the shareholders' benefit. But they are told to adhere to a process stan dard. in turn,designed to provide a circumstantial guarantee that they keep their eyes fi xed on the objective of shareholder value enhancement. 466 CARDOZO LAW REVIEW [Vol. 19:409 odic, documented confirmation that reinvested funds are devoted only to r > k projects? To what extent do management-instituted decision systems respect the integrity of capital budgeting method ologies? Are the results of such analyses included in board presentations of business plans? Do boards make dividend and re investment decisions independently of one another? Presumably, answers to these questions could be obtained through the mandatory disclosure system. Stepped up disclosure rules have been suggested in legal liter ature as a device for delimiting management discretion over the dividend. But, in making this argument, Victor Brudney rejected a requirement of detailed disclosure of the components underlying each dividend decision-"a dubious benefit, andat a likely intoler able cost. "169 He instead recommended that alterations in a longer term payout pattern be announced and explained. He sought, by reference to the signalling literature, to deploy the disclosure man date to clear up the ambiguities that attend departure from (or ad herence to) the conventional payout pattern.170 He acknowledged that his proposed disclosure rule might also constrain manage ment's self-interested tendencies respecting reinvestment, but only as an incidental benefit.1 71 Subsequent economic literature, particularly the agency expla nation of the dividend payout pattern, suggests that the gravamen of any disclosure problem lies on the reinvestment side of the coin. Indeed, Joseph Stiglitz recently suggested just such an extension of the mandate.172 And, certainly, if a disclosure mandate could be shaped that would discourage r < k investing without imposing an undue cost or litigation exposure burden on reporting firms, then such a mandate should be implemented. The question, then, is technical. Let us consider a few types of disclosure rules that could ad dress the problem, none of which would entail direct disclosure of investment projections. One type would require firms to report on their general policy respecting reinvestment decisions. They could state, for example, whether r < k investing is a conscious practice; they could state whether a cost-benetl.t analysis keyed to projected return and the cost of capital is an invariant requirement respecting WJ Brudney. supra note 32. at 116-17. 170 He placed primary emphasis on the dividend decision's apparent capacity to inllu encc the stock price. independent of the reinvestment decision. See id. at 117-22. 171 See id. at 122. 1 72 See EJlin & Stiglitz. supra note 111. at 1309. 1997] DIVIDENDS 467 capital investment decisions, and could state also whether particu lars respecting such analyses are reported and discussed at the board level; and they could report on their capital budgeting meth odologies and prevailing assumptions respecting the cost of equity capital. The problem is that any such talk still is cheap. Even with state of the art capital budgeting methodology and board level re porting, r < k investment can remain a habit so long as the system provides an input of inflated projected returns. In the alternative (or in addition), a second strategy could be employed. Firms could be required, in accordance with Stiglitz's suggestion, to identify the different investment projects adopted and funded in a given period and state the amount invested. Stig litz argues that this requirement would ameliorate a problem of perverse incentives-rational managers will invest in suboptimal projects where their success or failure is likely to be obscured by the property of unobservability.173 The problem with his sugges tion lies in the cost of creation and implementation. After all, it contemplates a whole new reporting system. If suboptimal investment practice really is a serious problem, then the cost-benefit question respecting these suggestions (or other disclosure strategies that might be brought forward) remains open. One anticipates the objection that additional disclosure re quirements amount to a deadweight cost. But that shopworn point dates from the era when managers were subject to an active control market deterrent. It lost vitality when the disciplinary equilibrium of the securities and control markets shifted radically in the late 1980s. The better contra argument today is that disclosure problems should be treated with contractual solutions rather than with government mandates. Disclosure rules are, after all, con tractible to some extent, and contractual commitments in theory present a superior alternative to state mandates. Unfortunately, however, practices of institutional investor intervention have not yet evolved to the point of active exploration of the corporate char ter's potential to contain more productive terms for contractible subject matter. As noted above,174 state law erects a barrier to any such development. Accordingly, a regulatory inquiry into the fe a sibility of disclosure conventions that import greater transparency respecting dividend and investment policy would by itself be wel come, if only because it could prompt private sector movement in this direction. 173 See irl. at 1301-02, 1307, 1309. 174 See supra text accompanying note 161. 468 CARDOZO LAW REVIEW [Vol. 19:409 CoNCLUSION If divid�nds are a puzzle because good firmspay them despite the availability of exceptional investment opportunities, then War ren Buffett is our only rational corporate manager because his pur suit of exceptionally good investments leads him to withhold dividends entirely. But, without any diminution in our recognition of his achievements, we can go some distance in solving the divi dend puzzle by reference to the economics of noncontractibility. The solution of course is that there is no neat solution, due to a confluence of present information asymmetries, imperfect incen tives, and future uncertainties that surround every dividend deci sion. We accordingly should reconceive dividend policy to be less a puzzle than a complex phenomenon observed among actors deal ing with one another in a second-best world. Thus reconceived, dividend policy admits of a complex explanation, each component of which is unsatisfactory but the entirety of which provides a satis factory working picture. The models applied in this Article add much detail to that working picture without suggesting any radical revision of its basic outlines. As such, they reconfirm the standing view that the dividend is a territory unsuited to the imposition of new legal mandates, even a mandate transferring declaration au thority over to the shareholders. Only as to disclosure policy might we plausibly explore productive law reform initiatives. Meanwhile, shareholders themselves, led by activist investment institutions, have found ways to make suboptimal earnings retention a less likely event, if not to guarantee against it. An empirical question arises respecting the extent of their success. Has the management climate of the 1990s so changed as to make suboptimal earnings retention a secondary problem on the corporate landscape? One suspects that once data are on the table for inspection, the answer still will lie in the judgment of the observer. 1997] DIVIDENDS 469 APPENDIX In the first quarter of 1996, Berkshire Hathaway held signifi cant stakes in the following publicly-traded companies: American Express, Walt Disney, Coca-Cola, Federal Home Loan Mortgage Corp., GEICO, Gillette, Wells Fargo & Co., and Salomon Broth ersY5 The chart176 that follows shows the dividends per share and earnings per share for each of these holdings along with those of companies in the Fortune 500 that fall into the same industry group for each of 1994 and 1995. For each company, a "payout ratio" is calculated in terms of dividends as a percentage of earnings. In addition the mean "payout ratio" is calculated for each industry group. The payout ratios of the Berkshire-held companies and the in dustry groups compare as follows: 1995 1994 Coca-Cola 37.29% 39.39% Beverages 53.62 % 51.93% Salomon Bros 18.29% -14.85% (loss; dividend paid) Brokerage 34.79% 38.26% Gillette 32.43% 31.85% Metals 10.55% 26.01 % Wells Fargo 23.12% 27.74% Banks 45.79% 41.45% Amex 28.94 % 31.64% FHLM 21.09 % 17.93% Div. Fin. 26.24% 28.61% Disney 13.46% 14.22% Ent. -7.56% -6.98% (due to Time Warner losses) GEICO 36.36% 26.11% Insurance 17.85% 14.8 1 75 See B ERKS H IRE HATHAW;\Y It'c.. 1995 A NNU A L REPORT 15-16 (1996). As of De cember 31. 1995. the Walt Disney holding was in the form of a long-term stake in Capital Cities/ABC Effective as of January 1996. Capital Cities merged with Walt Disney. Pursu ant to the merger. Capital Cities shares were converted into Walt Disney shares. Also. in 1996 Berkshire purchased the remaining 49'Yo of GElCO to make it a 100% owned subsidiary. ! 76 The chart was compiled from the following sources: Th e Fo rtune 1000 Ranked Wi thi11 Industry. foRTUNE. Apr. 29. 1996, at F-43-f-64; Company Ticker Symbol and All Share Information. Bloomberg On-Line financial Services, Sept. 20. 1996. � _p.. TICKER 1995 PAYOUT 1994 PAYO UT ---.j 0 COMPANY NAME SYMBOL DPS EPS RATIO DPS EPS RATIO BEVERAGES (7) Coca-Cola KO $0.44 $1.18 37.29% $0.39 $0.99 39.39% Anheuser-Busch BUD $0.84 $1.25 67.20% $0.74 $1 .94 38.14% Coca-Cola Enterprises CCE $0.50 $0.62 80.65% $0.50 $0.52 96.15% $0.97 Whitman WH $0.37 $1.26 29.37% $0.33 34.02% Q AV ERAGE $0.54 $1.08 53.62% $0.49 $1.:1 1 51.93% >:1 t::J a BROKERAGE (7) N $1.01 $5.42 18.63% $0.89 $4.74 Merrill Ly nch MER 18.78% a Lehman Bros LEH $0.20 $1.76 11.36% $0.18 $0.69 26.09% t--< $0.64 $3.50 18.29% $0.64 ($4.31) -14.85% � Salomon Bros SB � Paine We bber PWJ $0.48 $0.52 92.31% $0.58 $0.41 141.46% $0.54 $1.62 33.33% $0.52 $2.62 19.85% >:1 Bear Stearns BSC tr-1 AV ERAGE $0.57 $2.56 34.79% $0.56 $0.83 38.26% :s tr-1 � META L PRODUCTS (16) Gillette G $0.60 $1.85 32.43 % $0.50 $1.57 31.85% Crown Cork & Seal CCK $0.00 $0.83 0.00% $0.00 $1.47 0.00% MAS CO MAS $0.73 ($2.77) -26.35% $0.69 $1.22 56.56% � Ty co International TYC **$0.20 $1.41 14.18% $0.20 $1.28 15.63% - Illinois To ol Works ITW $0.64 $3.29 19.45 % $0.54 $2.45 22.04% !----" \0 U.S. I ncl ust ries USN N/A* N/A N/A N/A* N/A N/A :::;. 0 Stanley Wo rks SWK $0.71 $0.67 105.97% $0.69 $1.40 49.29% \0 ,_... Ball BLL $0.60 ($0.72) -83.33% $0.60 $2.20 27.27 % \0 \0 Newell NWL $0.46 $1.41 32.62% $0.39 $1.24 31.45% -....] ...... TOTA L AV ERAGE $0.44 $0.66 10.55 % $0.40 $1.43 26.01 % AV ERAG E W/OUT UNAVAILABLES ( *) $0.49 $0.75 11.87% $0.45 $1.60 29.26 % ** JUNE FISCAL YEAR ENDS COMMERCIAL BANKS (55) Ci ti corp CCI $1.20 $6.48 18.52 % $0.45 $6.29 7.15% Bankamerica BAC $1.84 $6.45 28.53% $1.60 $5.33 30.02% Nations Bank NB $2.08 $7.04 29.55% $1.88 $6.06 31.02% Chemical CHL $1. 88 $6.47 29.06 % $1.58 $4.54 34.80% $3.00 $6.36 47. 17% $2.72 $6.02 45.18% J.P. Morgan JPM b Chase Manhattan CMB $1.88 $6.04 31.13% $1.58 $4.97 31.79% ...... First Chicago NBD FCN $1.32 $3.41 38.71 % $1.17 $3.58 32.68% :s Fi rst Union Corp. FTU $1.96 $5.04 38.89% $1.72 $4.58 37.55% b Bane One ONE $1.21 $2.91 41.58% $1.09 $2.20 49.55% t'l'1 Bankers Tr u st BT $4.00 $2.03 197.04% $3.60 $7.17 50.21% 6 Fl eet Financial FLT $1.60 $1.57 101 .91% $1.30 $3.09 42.07% V") Norwest NOB $0.90 $2.73 32.97 % $0.77 $2.41 31.95 % PNC Bank PNC $1.40 $1.19 117.65% $1.31 $2.52 51.98% Keycorp KEY $1.44 $3.45 41.74% $1.28 $3.45 37.10% Bank of Boston BKB $1.29 $4.43 29.12% $0.93 $3.61 25.76% Wells Fargo WFC $4.60 $19.90 23.12% $4.00 $14.42 27.74% Bank of New Yo rk BK $0.68 $2.15 31.63 % $0.53 $1.85 28.65 % First Interstate Bancorp I $3.10 $11.02 28.13% $2.75 $2.31 119.05% Mellon Bank MEL $2.00 $4.46 44.84 % $1.56 $2.42 64.46% Wachovia WB $1.38 $3.49 39.54% $1.23 $3.12 39.42 % Suntrust Banks STI $0.74 $2.47 29.96 % $0.66 $2.18 30.28 % Barnett Banks BBI $0.88 $2.57 34.24% $0.77 $2.33 33.05% +>- -....] National City NCC $1.30 $2.95 44.07% $1.18 $2.70 43.70% ,_... ' , ''·''i'·-.. 1 � First Bank System FBS $1.45 $4.11 35.28% $1.16 $2.14 54.21% --...) N Comerica CMA $1.34 $3.52 38.07 % $1.20 $3.28 36.59% Boatmen's Bancshares BOAT $1.39 $3.25 42.77% $1.25 $3.17 39.43 % U.S. Bancorp USBC $1.03 $2.09 49.28 % $0.91 $1.60 56.88% Co restates CFL $1.36 $3.18 42.77% $1.20 $1.72 69.77% Republic New York RNB $1.44 $4.59 31.37% $1.26 $5.61 22.46% MBNA KRB $0.54 $1.54 35.06% $0.46 $1.18 38.98% TOTA L AVERAGE $1.67 $4.56 45.79% $1.44 $3.86 41.45% Q DIVERSIFIED FINANCIAL (15) ::::0 tJ Fed. Nat'! Mortgage Assoc. FNM $0.68 $1.95 34.87 % $0.60 $1.94 30.93% $0.90 $3.11 28.94 % $0.87 $2.75 31.64% () Am erican Express AXP N Morgan Stanley Group MGMS $0.48 $3.33 14.41% $0.60 $2.09 28.71% () $1.20 $5.69 21.09% $1.04 $5.80 17.93 % Fed. Home Loan Mortgage FRE l""< Dean Witter DWD $0.61 $4.88 12.50% $0.48 $4.27 11.24% � American General AGC $1.24 $2.64 46.97 % $1.16 $2.45 47.35% � Household International HI $1.29 $4.30 30.00 % $1.21 $3.50 34.57% ::::0 Berkshire Hathaway BRK $(l.00 $610.90 0.00% $0.00 $420. 12 0.00 % � Student Loan Mktg. Assoc. SLM $1.51 $7.20 20.97 % $1.40 $4.91 28.51% :s $2.91 $5.53 52.62 % $2.79 $5.05 55.25% � March & McLennan MMC � TOTA L AV ERAGE $1.08 $64.95 26.24'Yo $1.02 $45.29 28.61 % ENTER'T'AINMENT (4) Wall Disney DIS $0.35 $2.60 13.46% $0.29 $2.04 14.22% Via com $0.00 $0.43 0.00% $0.00 $0.07 0.00% � VIA ...... $0.36 ($0.57) -63.16% $0.35 ($0.27) -129.63% Time-Warner TWX I-' Turner Broadcasting TBS $0.07 $0.36 19.44% $0.07 $0.08 87.50% \0 � AV ERAG E $0.20 $0.71 -7.56% $0. 18 $0.48 -6.98% 0 \0 f-.-" INSURANCE, P&C, STOCK (27) \0 \0 American Int'l Group AIG $0.32 $5.30 6.04% $0.28 $4.58 6.11% --....) ....__.. Allstate ALL $0.78 $4.22 18.48% $0.72 $1.08 66.67% Loews LT R $0.62 $14.98 4.14% $0.50 $2.22 22.52% Travelers Group TRY $0.54 $3.67 14.71% $0.43 $2.57 16.73% ITT Hartford Group HIG $0.00 $4.77 0.00% $0.00 $5.50 0.00% General RE GRN $1.96 $9.92 19.76% $1.92 $7.97 24.09% Chubb CB $0.98 $3.92 25.00% $0.90 $2.97 30.30% St. Paul Cos. SPC $1.58 $5.68 27.82 % $1.49 $4.93 30.22% Sareco SAFC $1.02 $3.15 32.38% $0.95 $2.48 38.31% American Fi n. Group AFG $0.75 $3.88 19.33% $1.16 ($0.83) -139.76% USF&G FG $0.20 $1.53 13.07% $0.20 $1.77 11.30% Allmerica Financial AFC $0.00 $2.61 0.00% $0.00 $0.76 0.00% t::::1 ...... GEICO GEC $1.08 $2.97 36.36% $1.00 $3.83 26.11% $0.22 $3.24 6.79% $0.21 $3.59 5.85% :s Progressive PGR t::::1 $0.32 $0.73 43.84% $0.32 $0.38 84.21% Reliance Group Holdings REL 1:11 $0.69 $4.70 17.85% $0.67 $2.92 14.84% < AV ERAGE t::::1 V) +:>. --....) w